Issue 11647: mismatch between diagram
Issue 12165: URGENT SBVR.xsd issue
Issue 12437: Issue "fact type role is in fact type"
Issue 12531: editorial issue -- example is missing a line
Issue 12540: Clause 8 does not include the concepts needed to represent itself
Issue 12541: No relationship defined between clause 8 concepts and clause 11 concepts
Issue 12542: terminological dictionary
Issue 12543: A rulebook should have a URI
Issue 12589: "characteristic type" should be a "category type"
Issue 12614: SBVR typos
Issue 12849: fact type 'fact type form incorporates fact symbol' needs additional captio
Issue 12956: Note for individual concept does not follow from the Definition
Issue 13135: SBVR Issue: can a role range over multiple object types
Issue 13138: Move Fact Model Container Concepts from Clause 8 to Clause 10 (Spin-off from Issue 12540)
Issue 13139: Clean up and Complete Vocabulary for Clause 10.1.1
Issue 13150: Issue: SBVR Clause 8
Issue 13716: Definitions in subsection 11.1.5
Issue 13802: SBVR Issue: What is a fact type form
Issue 13803: SBVR Issue: Definition of signifier
Issue 13804: SBVR Issue: Model expression structure
Issue 13835: Use of the Signifier "Fact Model"
Issue 13836: Note for Is-Facet-of Fact (Type)
Issue 13849: SBVR did not pick up the ISO synonym "Part-Whole Relation
Issue 13850: The segmentation 'Thing in Context' is inconsistent with the definitions of 'role' and 'facet'
Issue 13851: Definition of Is-Property-Of Fact Type
Issue 13865: SBVR Issue : Inconsistent use/definition of keyword 'or'
Issue 13996: SBVR Fig 12-1 tweak
Issue 14029: Conflation of Proposition with "Proposition + Performative " plus Disconnect between Concept and Proposition
Issue 14241: Coexistence approach to SBVR
Issue 14843: Concepts-centric Model and Fact Model are different
Issue 14844: Move the Definitions in Subclause 8.5 to Clause 13
Issue 14849: Instances of Clause 8 fact type should be states of affairs
Issue 15008: Use of "denotes" in note for "state of affairs"
Issue 15124: Inconsistent use of terminology when relating facts to fact types
Issue 15151: new SBVR issue - relationship of 'vocabulary' and 'rulebook'
Issue 15153: New SBVR Issue: "Template" & "Templating
Issue 15157: Existential and Elementary
Issue 15250: SBVR - change to Definition of 'fact type'
Issue 15314: Definition of Vocabulary
Issue 15402: No normative reference to ISO 6093
Issue 15403: 'quantity' and 'number' are not formal logic concepts
Issue 15404: Set requires distinguished things
Issue 15450: [SBVR] fact type role designation
Issue 15623: "The Signifier "Fact Type" Badly Misrepresents the Clause 8.1.1 Concept as Defined and Needs to be Replaced"
Issue 15635: Placeholder concepts model SBVR Structured English syntax
Issue 15684: SBVR recognizes the notion of "property" in Clause 11.1.5 in "is-property-of", but never defines the concept directly
Issue 15805: SBVR editorial issue
Issue 15837: Error in Example for "noun concept nominalization"
Issue 15840: SBVR - Error in MeaningAndRepresentation-Model.xml
Issue 15841: SBVR Editorial Issue - closed projection defines noun concept
Issue 15947: Inconsistency in is-role-of and is-category-of fact types
Issue 15948: is-property-of fact types
Issue 15949: assortment fact types
Issue 15950: inappropriate definitions of burinsss rule, rule statement
Issue 15951: example definitions (of "Australian")
Issue 15952: example elementary fact
Issue 15953: 'reality' and 'in-practice' models
Issue 15972: Example of quantity vs. quantification
Issue 16020: Individual Concept and Change
Issue 16059: Governed Community & Adoption of Business Rules
Issue 16062: SBVR Issue: Move 'rulebook'
Issue 16101: Explicitness of Representation
Issue 16103: Conflation of the signifier “rulebook” with the concept/definition for Speech Community Representations
Issue 16166: Distinguishing between Representation Expressions With and Without Embedded Markup
Issue 16171: SBVR typo - p. 26
Issue 16172: Clarify difference between EXISTS and OCCURS
Issue 16258: A statement may express no proposition
Issue 16309: Clarify Objectification
Issue 16314: SBVR issue: Can there be multiple instances of a thing?
Issue 16375: Adoption of Concepts
Issue 16486: SBVR Issue - Relationships between States of Affairs
Issue 16491: "Objectification" Needs to Be Renamed
Issue 16522: "Nominalization" Needs to Be Renamed
Issue 16523: "Aggregation Formulation" Needs to Be Adjusted
Issue 16524: "Projection" Needs to Be Renamed
Issue 16525: "Quantification" Needs to Be Renamed
Issue 16526: Definition of proposition
Issue 16527: SBVR ISSUE - definite description
Issue 16555: 'Variable' should be renamed as 'formulaic variable' or its meaning clarified
Issue 16610: SBVR issue - Need verb concept to support "local closure"
Issue 16630: Actuality demonstrates Proposition
Issue 16631: The formal logic interpretation for SBVR in Common Logic (CL) given in Clause 10 is incomplete
Issue 16683: Define that Clause 10 ‘Fact Models’ are by Default Closed World Models
Issue 16684: SBVR Vocabularies Relationship to SBVR Subclause 10.1.1
Issue 16685: Fix Entries in Subclause 10.1.2.1 to Align with Subclause 10.1
Issue 16727: "thing has property".
Issue 16913: Urgent issue on SBVR 1.1 RTF (NOT SBVR 1.2)
Issue 17017: SBVR makes use of ElementImports to give additional aliases to some elements in the same package
Issue 17068: Simplification of presentation of Annex E
Issue 17097: SBVR issue - Re-sequencing Clause 11
Issue 17098: "Three Editing Instructions Overlooked in Issue 17017 Resolution
Issue 17144: typo in clause 10.1
Issue 17241: Annex H recommends faulty UML constructs
Issue 17243: Precedence of operators
Issue 17244: Keyword "another"
Issue 17269: Use of morphological variants of terms is inadequately addressed
Issue 17414: Clarify Purpose and Scope of SBVR, the Authority for SBVR Vocabulary Content, and SBVR Vocabularies
Issue 17439: Individual Verb Concept
Issue 17440: Redefinition of "Body of Shared Concepts" (Clause 11)
Issue 17441: Definition of "representation uses vocabulary" (Clause 11
Issue 17452: New SBVR issue - Re-sequencing Clause 8
Issue 17527: Correct ambiguities in signifiers and definitions of noun concepts
Issue 17532: Noun form designates two different concepts
Issue 17542: Inadequate, Overlapping and Disorganized Specs for Sets and Collections of Concepts, Meanings, and Representations
Issue 17571: Distinguishing the senses of infinitive and present tense
Issue 17599: Align Definitions of Modal Entries in Clauses 8, 9 & 10
Issue 17791: How can an attributive role be declared?
Issue 17792: Clause 10.1.2 Vocabulary Clarifications
Issue 17819: Scope of an SBVR Body of Shared Concepts
Issue 18166: individual verb concept’ in SBVR
Issue 18172: Add Generic Occurrence to SBVR to Support Other Specifications for Occurrence in Time, Space or Other Dimensions
Issue 11647: mismatch between diagram (sbvr-rtf)
Click here for this issue's archive.
Nature: Revision
Severity: Minor
Summary: mismatch between diagram where speech community is associated with exactly one semantic community but 07-09-04 version of the XMI/CMOF has speech community mapping to multiple semantic community e.g. <ownedMember xmi:type="cmof:Association" name="semantic community has speech community" xmi:id="semanticCommunityHasSpeechCommunity" memberEnd="semanticCommunityHasSpeechCommunity.semanticCommunity semanticCommunityHasSpeechCommunity.speechCommunity"> <ownedEnd xmi:type="cmof:Property" name="semantic community" xmi:id="semanticCommunityHasSpeechCommunity.semanticCommunity" type="semanticCommunity" lower="0" upper="*"/> <ownedEnd xmi:type="cmof:Property" name="speech community" xmi:id="semanticCommunityHasSpeechCommunity.speechCommunity" type="speechCommunity" lower="0" upper="*"/> </ownedMember>
Resolution:
Revised Text:
Actions taken:
November 12, 2007: received issue
Issue 12165: URGENT SBVR.xsd issue (sbvr-rtf)
Click here for this issue's archive.
Click here for this issue's attachments.
Source: Chronolytics (Mr. David Carlson, dave(at)chronolytics.com)
Nature: Revision
Severity: Critical
Summary:
The final XMI Schema for SBVR serialization is not correct for Associations, as required by the XMI 2.1.1 specification. An implementation that produces a valid XMI serialization will be judged as invalid, according to the SBVR.xsd. This is a critical bug. I have created an SBVR implementation using Eclipse EMF, based on the final SBVR cmof model. An example model serialization from EMF is attached, as test.sbvr. In it, each model element includes an xmi:id attribute. However, the SBVR.xsd does not allow this id on types derived from cmof Association. >From XMI v2.1.1, p. 49, the AssnAtts must include all XMIFixedAttribs 7. AssociationDef ::= "<xsd:element name='"' 7a:AssnElmtName '"'>" "<xsd:complexType> <xsd:choice minOccurs='0' maxOccurs='unbounded'>" 7b:AssnContents "</xsd:choice>" 7d:AssnAtts "</xsd:complexType> </xsd:element>" 7a. AssnElmtName ::= 1c:Namespace //Name of association// 7b. AssnContents ::= 7c:AssnEndDef 7c:AssnEndDef 4c:Extension 7c. AssnEndDef ::= "<xsd:element" "name='" //Name of association end// "'>" "<xsd:complexType>" 1g:XMIFixedAttribs "</xsd:complexType>" "</xsd:element>" 7d. AssnAtts ::= 1g:XMIFixedAttribs And, from p. 44, the XMIFixedAttribs 1g. XMIFixedAttribs ::= ( "<xsd:attribute ref='xmi:id'" "use='optional'>" | "<attribute name='" //Id attrib name// "'" "type='xsd:ID' use='optional'") "<xsd:attributeGroup ref='xmi:ObjectAttribs'/>"
In clause 8.1.1.1, we have "fact type has role", with a synonymous form "fact type role is in fact type". Figure 8.2 also shows "fact type role is in fact type". Issue: a "fact type role" is a specialization of "role", so it is confusing to see the preferred form of the fact type use "role" but the synonymous form use "fact type role". Especially because figure 8.2 seems to indicate that a "fact type role" is in a fact type but that a "role" is explicitly *not* in a fact type. So the figure appears to contradict "fact type has role". Recommendation: I think the preferred entry is wrong, and should be changed to "fact type has fact type role".
In section 9.2.8, on page 70, the example for "aggregation formulation" introduces several variables. All but one of the introduced variables is specifed as ranging over some concept. For example, ". . . . The second variable ranges over the concept ‘number’." My issue: there is no corresponding "ranges over" line for the third variable. It is true (per 9.2.1) that variables need not range over any concept. But this example would be clearer if the "ranges over" line were included for that third variable. I believe this third variable is supposed to range over the concept 'set'.
SBVR currently has multiple concepts for organizing vocabularies and rules:
* conceptual schema (clause 8.5)
* fact model (8.5)
* body of shared meanings (11.1.1)
* body of shared concepts (11.1.1)
* terminological dictionary (11.1.1)
* vocabulary (11.1.1)
* rulebook (11.2.2.4)
Some issues:
1) Clause 8 does not include the concepts needed to represent itself, even
though clause 2 says clause 8 is a standalone compliance point. Clause 8
claims to be a vocabulary, but the concept "vocabulary" is in clause 11,
not clause 8. Hence an implementation of clause 8 cannot model clause 8
itself.
SBVR currently has multiple concepts for organizing vocabularies and rules:
* conceptual schema (clause 8.5)
* fact model (8.5)
* body of shared meanings (11.1.1)
* body of shared concepts (11.1.1)
* terminological dictionary (11.1.1)
* vocabulary (11.1.1)
* rulebook (11.2.2.4)
Some issues: 2) No relationship is defined between the clause 8 concepts and the clause
11 concepts. Is a body of shared concepts based on a conceptual schema?
How does a fact model relate to a terminological dictionary?
SBVR currently has multiple concepts for organizing vocabularies and rules:
* conceptual schema (clause 8.5)
* fact model (8.5)
* body of shared meanings (11.1.1)
* body of shared concepts (11.1.1)
* terminological dictionary (11.1.1)
* vocabulary (11.1.1)
* rulebook (11.2.2.4)
Some issues:
3) A terminological dictionary should be able to incorporate other
terminological dictionaries, as with "vocabulary incorporates vocabulary".
Otherwise, we cannot structure terminological dictionaries in parallel with
vocabulariesSBVR currently has multiple concepts for organizing vocabularies and rules:
* conceptual schema (clause 8.5)
* fact model (8.5)
* body of shared meanings (11.1.1)
* body of shared concepts (11.1.1)
* terminological dictionary (11.1.1)
* vocabulary (11.1.1)
* rulebook (11.2.2.4)
Some issues:
4) A rulebook should have a URI, so that the rulebook can be addressed over
the Internet.Section 11.1.2.2 "Kinds of Characteristic" on page 136 says that "characteristic type" is "General Concept: concept type". I suggest that "General Concept: categorization type" would be more accurate. Given this proposal, in EU-Rent, making "branch type" a "characteristic type" would enable statements such as "if there exists a branch that is a city branch ...."
attached is a dcument containing SBVR typos
p. 150 (PDF p. 162), Clause 11.2.1.2, to the entry for 'fact type form incorporates fact symbol': Add the following caption, to appear after the current Synonymous Form caption: Synonymous Form: fact type form demonstrates designation using term styling where underlined (above) and verb styling for italics (above) Also, on this same page, there is a typo in the Definition caption under the entry for 'fact symbol': In 'fact type form' (which ends the Definition) the first space needs to be underlined — i.e., apply term styling to the space.
Clause 8.1.1 Concept: individual concept The Definition of 'individual concept' is: concept that corresponds to only one object [thing] The Note says: "each referring individual concept has exactly one and the same instance in all possible worlds" "Corresponds to only one object" (in any possible world) is not at all the same thing as "corresponds to exactly one and the same object in all possible worlds". One of the definition and the Note should be corrected. I would prefer changing the definition to match the note. Note also that changing the definition means that "the President of the United States" is an 'individual concept' that denotes an office, but not a person. And the concept "the person who is President of the United States" is _not_ an 'individual concept'.
The fact type "role ranges over object type " appears in section 8.1.1. As defined -- due to the "open world" aspect of SBVR -- it appears that a role can range over multiple object types, which does not make much sense. But if you look at the MeaningAndRepresentation-model.xml file, you will find confirmation that a role can range over multiple object types. This has a downstream impact in the MDT-SBVR open source Eclipse project, where the .xml file is converted directly to an EMF model and a matching Java implementation. The API for setting an instance of this fact type permits each role to range over multiple object types. This has two impacts: (a) adds complexity to the API; (b) forces tool vendors to try to figure out the semantics of one role that ranges over multiple object types. Either the specification should explain what it means for a role to range over multiple object types, or it should introduce a Necessity: "each role ranges over exactly one object type".
Please see attached Word document for Issue details. This SBVR spin-off Issue is a part of a package of three proposed Issue resolutions: - the proposed resolution of this spin-off Issue which will be posted when it has a number; - the proposed resolution to Issue 12540; and - the proposed resolution of the 11296-1a / 11303-b spin-off Issue which will be posted when it has a number.
SBVR Issue -- Clean up and Complete Vocabulary for Clause 10.1.1 (Was Issues 11296-1a and 11303-b) (Part of Separating 11296 & 11303 into Manageable Pieces)Please see attached Word document for Issue details. This SBVR spin-off Issue is a part of a package of three proposed Issue resolutions: - the proposed resolution of this spin-off Issue which will be posted when it has a number; - the proposed resolution to Issue 12540; and - the proposed resolution of the Issue 12540 spin-off Issue which will be posted when it has a number.
understand you will be discussing the topic of packaging SBVR tomorrow, and I want to provide a perspective on this topic and make a request. In my view, the key packaging concepts “fact model” and “conceptual schema” need to be in the normative SBVR metamodel to support widespread sharing and reuse of SBVR models. We want to promote the development of libraries of SBVR fact models and conceptual schemas and to compose fact models and conceptual schemas from other fact models and conceptual schemas. The ability to package these in a standard way is crucial to this end. A normative approach to globally identifying these models is needed to support their sharing and reuse. Concepts of packaging, identification, and composition of fact models and conceptual schemas are preferably included in Clause 8. As the most basic compliance point, Clause 8 needs to be expressible in terms of itself, and to include concepts for packaging, identification, and composition of fact models and conceptual schemas. I understand a proposal is under consideration to move “fact model” and “conceptual schema” entries to Clause 10. This would be a mistake, as we would then have no normative way of specifying the packaging. The definition of “conceptual schema” should be refined to reflect the fact that a conceptual schema is a kind of fact model. The distinction between a conceptual schema and other fact models is that a conceptual schema includes at least one fact that asserts the existence of a concept. Other fact models that are not conceptual schemas contain only ground facts. The text of SBVR makes it clear that a conceptual schema is a fact model, that every SBVR interchange document is a fact model. That “conceptual schema” specializes “fact model” should be reflected in the definition of “conceptual schema.” The term “vocabulary” is not used in the SBVR specification consistently with its definition as a “set of designations and fact type forms…” Each of the normative clauses of SBVR, called a “Vocabulary,” is actually an annotated conceptual schema. A conceptual schema comprises a “combination of concepts and facts (with semantic formulations that define them)…” The designations and fact type forms in each SBVR normative “Vocabulary” constitute the vocabulary of that “Vocabulary”. The definitions and necessities in the SBVR entries are statements of schema facts. The notes and examples are annotations of the conceptual schema. Ability to include annotations is crucial to practical development and use of any model, and is universally provided for in other and modeling and programming languages. It should be possible to normatively include annotations in a SBVR conceptual schema or fact model. Accordingly, it is recommended that “description” and related concepts of notes and examples in Clause 11.2.2 be moved to Clause 8 to support annotation of fact models. With respect to the semantic formulations of a conceptual schema, it is preferred that Clause 8 only include statements of the definitions and schema facts, and leave it to Clause 9 to include the semantic formulations of these. Either “vocabulary namespace” and fact types that use the term should be moved to Clause 11, or “vocabulary” should be moved to Clause 8. The concept “vocabulary” is not necessary in Clause 8 but might be conveniently located there. Namespaces adequately serve the purpose of organizing designations and fact type forms. It is suggested the RTF consider providing recommendations for naming conventions for URIs of namespaces and related conceptual schemas that define and constrain the concepts represented by the designations and fact type forms in the namespaces. Here are some suggested entries for Clause 8 that show how the concepts described above might be modeled: conceptual schema Definition: fact model that includes at least one existential fact asserting a concept Note: This definition extends the definition of ‘conceptual schema’ in SBVR to formalize that a conceptual schema is a kind of fact model. This is evident in the specification text, but is not included in the current definition. Note: The facts of a conceptual schema in addition to the concept existential facts describe what is possible, necessary, permissible, and obligatory in each possible world of the domain being modeled. Note: Two levels of formalization of fact models (including conceptual schemas) are possible. 1) A fact model may contain only statements of definitions and other facts and not their semantic formulations. In this case, the fact model can meet the Meaning and Representation compliance point, 2.2.1. 2) A fact model may contain semantic formulations of its definitions and facts, in which case the fact model can meet the Logical Formulation of Semantics compliance point, 2.2.2. fact model1 includes fact model2 Note: This fact type enables recursive composition of fact models and conceptual schemas. Necessity: This fact type is reflexive, antisymmetric, and transitive, i.e. related fact models are at least partially ordered. fact model includes description Note: This fact type enables the annotation of fact models and conceptual schemas. thing has URI Note: This fact type enables modeled things to be identified globally for future reference. I am requesting that these concepts, or some refinement of them, be included in the next release of SBVR.
A number of the definitions in this subsection are incomprehensible, and not well integrated with the rest of the SBVR vocabulary. These definitions center on: assortment fact type, categorization fact type, is-role-of fact type, and is-facet-of fact type. Also, these concepts are defined as kinds of "fact types", but should actually be defined as kinds of *facts*. Finally, the order of the entries needs adjustment as a result of the above.
Title: What is a fact type form? Specification: SBVR Version: 1.0 Source: Ed Barkmeyer, NIST, edbark@nist.gov Summary: In SBVR, clause 8.3.4, 'fact type form' has the definition: "representation of a fact type by a pattern or template of expressions based on the fact type". According to clause 8.3(.0), 'representation' is "actuality that a given expression represents a given meaning". Is "a pattern or template of expressions" an "expression"? According to 8.2, a 'signifier' is "expression that is a linguistic unit or pattern [of sounds or symbols]". So apparently there are expressions that are patterns and they can be signifiers. Per 8.3.1, designation is the "representation of a concept by a sign", and a fact type is a concept, so it may have a representation that is a designation. But the UML diagram shows that a fact type form is not a designation. So presumably a 'pattern or template of expressions' is not a 'sign'. But a signifier, which is a pattern, must be a 'sign', because it is the expression that participates in a designation. But the expression of a fact type form is apparently not a signifier, since only designations have a 'signifier' role, and a fact type form is not a designation. The inconsistency in the terminology, and the failure to make clear parallels and distinctions, is very confusing. It seems that the idea here is that an 'expression' can be a structure of individual sub-expressions, and that, in representing a fact type, the structure and the sub-expressions play distinct roles in the "actuality" of representing the fact type. This means that at least this idea of structured expressions should be described in clause 8.2, as a kind of expression more interesting than "text". It appears to be the intent that a fact type form expression always has a structure with representation sub-behaviors. Is that what distinguishes a fact-type form from a designation? The text is completely silent as to what the delimiting characteristic is. The remaining question then is: what kind of representation is exemplified in a terminological entry for a fact type in the SBVR vocabulary itself? E.g., is "designation has signifier" a designation for a fact type or a fact type form for it? (According to the UML diagram it cannot possibly be both.) And if the latter, does an SBVR fact type not actually have a designation? More confusion. Recommendation: 1. Define the concept that is "pattern or template of expressions" in 8.2 2. Use these structure concepts to define the nature of a fact type form in 8.3.4. For example, a placeholder is a sub-expression. 3. Specify the distinguishing characteristic of a fact-type form that makes it different from a designation. 4. Specify what the vocabulary entries for fact types are: fact-type forms or fact-type designations.
Title: Definition of signifier Specification: SBVR Version: 1.0 Source: Ed Barkmeyer, NIST, edbark@nist.gov Summary: SBVR clause 8.2 defines 'signifier' to be a role in a 'designation'. But the concept 'designation' is defined in 8.3.1. Recommendation: Move the entry for 'signifier' to 8.3.1, where it is used.
Title: Model expression structure Specification: SBVR Version: 1.0 Source: Ed Barkmeyer, NIST, edbark@nist.gov Summary: SBVR clause 8.2 defines 'starting character position' as a means of reference to a substring of a Text object. And the definition of placeholder in clause 8.3.4 treats the placeholder as a syntactic substring that is identified by its starting character position. This is a junior programmer model of expressions -- a poor PSM -- and it doesn't work reliably for a number of surface languages. The idea is that the unspecified representation of a concept may involve an expression that has a syntactic structure. Since SBVR has no idea what that syntactic structure is (because it belongs to an undefined surface language for which SBVR is the metamodel), it must define a general model of expressions sufficient to support the idea that a placeholder is a subexpression, and has a surface-language-defined means of identification. Recommendation: In 8.2, Delete 'starting character position'. Replace it with a model of expressions that makes clear the point at which surface-language grammar and orthography determine the technical structure of the expressions. In 8.3.4, delete all references to 'starting character position' in the entry for 'placeholder', and replace them with references to the structural concepts (to be) defined in 8.2. In 8.3.4, delete 'placeholder has starting character position' and replace it with a relationship to a structural concept (to be) defined in 8.2.
The signifier "fact model" should never be used in SBVR to include behavioral (deontic) elements of guidance. That usage makes no sense to business people, who would not expect anything labeled "fact[s]" to include rules. The origin of the idea meant by "fact model" and "conceptual model" predates any handling of deontic elements of guidance. In other words, deontic elements of guidances were not anticipated or treated by earlier approaches. We are just now catching up to the problem. The current definition of "fact model" (and "conceptual model") is: "combination of a conceptual schema and, for one possible world, a set of facts (defined by semantic formulations using only the concepts of the conceptual schema)". The resolution of this issue must involve at least the following: 1. Selection of a new signifier for the meaning expressed by the above definition. As a strawman, I would propose "Possible World Model". That sounds like something of concern to (only) tool engineers, which is appropriate, since the notion would not interest business people. 2. To suit the signifiers "fact model" and "conceptual model" the current definition must be modified to exclude facts pertaining to deontic elements of guidance. 3. All appearances of these signifiers in SBVR must be reviewed to determine which concept was actually meant. The meaning then given for the signifiers "fact model" and "conceptual model" is one that would be important to business people. If not significant for clause 8 (or 9 or 10), it can be moved to clause 11.
A Note for Is-Facet-of Fact (Type) currently reads: "A given community may choose to include only one facet." The Note could be read as a rule: It is permitted that a given community include only one facet." The Note should probably read: A given community may choose to include any number of facets, including just one or none at all.
The concept "Partitive Fact Type" is based on the Concept "Partitive Relation" in ISO 1087. However, SBVR did not pick up the ISO synonym "Part-Whole Relation". This could raise questions about how the SBVR notion is being based on the ISO notion. Also, "Part-Whole" is more business-friendly than "Partitive". Proposed Resolution: Add "Part-Whole Fact Type" as a synonym of "Partitive Fact Type". (If for some reason this is deemed inappropriate or undesirable, a note should be added as to why.)
The segmentation 'Thing in Context' is inconsistent with the definitions of 'role' and 'facet'. The segmentation is based on an assumption that the extensions of 'role' and 'facet' are completely disjoint. But there is nothing in the definitions of 'role' or 'facet' that cause them to be disjoint. It is possible that a situational role is relevant only from a certain viewpoint. Recommendation: Remove 'Thing in Context' and all references to it. Change Figure 11.1.5 to not show segmentation between 'role' and 'facet'.
The definition of is-property-of fact type is based on the notion of ‘essential quality’. Use of the word ‘essential’ is misleading since ISO and therefore SBVR talks about ‘essential characteristic’ in quite a different sense. The three Dictionary Bases are poorly chosen (probably because they were chosen before the ISO notion of characteristic was introduced into SBVR). In any event, the current definition of is-property-of fact type does not accurately express the intended meaning of the concept. Resolution: 1. Change the definition of "Is-Property-Of" fact type to: associative fact type that is defined with respect to a given concept such that each instance of the fact type is an actuality that an instance of the concept has a particular quality or trait 2. A better Dictionary Basis should replace the existing ones. Use the following definition from MWUD: 1 a : a quality or trait belonging to a person or thing;
Title: Inconsistent use/definition of keyword 'or'
Spec: SBVR
Version: 1.0
Source: Ed Barkmeyer, NIST, edbark@nist.gov
Summary:
In clause 9.2.1, p.52, 'bindable target' is defined as:
variable, expression or individual concept
In clause 11.1.5, 'contextualization fact type' is defined as:
is-role-of fact-type or is-facet-of fact-type
In clause 11.1.5, 'contextualized concept' is defined as:
role or facet
At the end of section C.3.2.1 in Annex C, the example is:
contextualized concept
Definition: role or facet
In Annex E, p.327, 'fuel level' is defined as:
full or 7/8 or 3/4 or 5/8 or 1/2 or 3/8 or 1/4 or 1/8 or empty
In all these, 'or' is stylized as a keyword. According to Annex C.3.2.1, these represent extensional definitions, i.e., the unions of the extensions of the concepts. But according to Annex C.1.1, the
keyword 'or' is defined to mean logical disjunction between two
propositions. So the definition of keyword 'or' is inconsistent with the usages.
One solution is to change the definitions.
E.g., for contextualized concept:
Definition: concept that is a role or is a facet
This form has a direct translation to the concepts in Clause 9.
An alternative is to change the meaning of the keyword in C.1.1, assuming it is never used for logical disjunction of propositions.
Another alternative is to introduce a new keyword.
Figure 12-1 shows 'merged' arrowheaded lines from 'element of guidance' and 'rule' into 'propositiion'. While this is not formally meaningful our graphics have used a convention to bring the lines together for elements that are mutually exclusive and to show the lines separate when not — ref. the separate lines into 'rule'. I suggest that Figure 12-1 be updated to show separate arrowheaded lines into 'proposition'.
There two closely related flaws in SBVR Clause 8.1:
1. a conflation of 'proposition' with "'performative' + 'proposition'"
2. a disconnect between 'concept' and its subcategories and 'proposition' and its subcategories which are really one concept or two perspectives on the same thing.
Conflation of 'Proposition' with "'Performative' + 'Proposition'"
- 'proposition' meaning that is true or false (the "semantic content"
part in 'proposition' + performative')
- 'proposition' + 'performative' (where the 'performative' part is the
"communicative function") e.g.:
o proposition + "deontic" performative = behavioral guidance
o proposition + "alethic" performative = definitional rule
o proposition + "taken to be true" performative = fact
The core meanings are in the propositions which are then made into something else by combination with a particular performative. This is why there is no reason to include the concept 'fact' at all in Clauses 8, 9 11 or 12 except to support the formulation of fact statements -- which are really out of scope for a standard for "concept(definition)-centric special purpose business language dictionaries plus guidance specifications in terms those definiton-centric dictionaries". Examples of general concepts can be provided by using names and fact type forms of individual concepts without needing to turn the individual concepts into facts (by adding the performative "taken to be true") so that fact statements can be used as examples.
Disconnect between 'Concept' and its Subcategories and 'Proposition' and its Subcategories
Clause 8.1 defines two concepts ('concept' and 'proposition') as if they were completely separate things when in fact they are at most two perspectives on the same thing:
· general noun concept = open (existential) proposition
· individual noun concept = closed (existential) proposition
· general verb concept = open (relational) proposition
· individual verb concept = closed (relational) proposition
(this is the verb concept that corresponds to a given state of affairs)
Resolution:
Remove the Conflation of 'Proposition' with "'Performative' + 'Proposition'"
1. Add the concept (definition) for "performative" and term it "communicative function" [3.7] as per ISO/CD 24617-2 "Language resource management -- Semantic annotation framework (SemAF) -- Part 2: Dialogue acts".
2. Add the three performative (communicative function) individual concepts used in SBVR: "taken to be true", "true by definition", and behavioral guidance.
3. Add the concept (definition) for "performative' + proposition" and term it "dialogue act" [3.2], as per ISO/CD 24617-2.
4. Show fact, behavioral guidance, and definitional guidance as concept type dialogue act with their respective performative (communicative function) instances instead of their current definition as subcategories of proposition.
5. Review all references to 'proposition' to determine whether the intended reference is to semantic content or to a discourse act (proposition + performative); e. g. statement expresses dialogue act (not proposition).
Remove the Disconnect between 'Concept' and its Subcategories and 'Proposition' and its Subcategories
1. Add open/closed proposition categories, and existential/relational proposition categories.
2. Fix the subcategories of concept to fit the above, and have both 'concept' and 'proposition' as more general concepts for the subcategories.
3. Replace all current uses of 'individual concept' to 'individual noun concept'.
Revised Text:
…to follow, including redrawn diagram(s)
According to our observations, more than 95% of all business applications operate under the closed world assumption and the state of affairs interpretation. In order to give other approaches (standards) the option to work with SBVR, it is proposed to offer the following: for each fact type one of the following combinations can be selected: 1. Closed world assumption; state of affairs interpretation 2. Closed world assumption; actuality interpretation 3. Open world assumption; state of affairs interpretation 4. Open world assumption; actuality interpretation. For convenience it is recommended to add the following four meta fact types: 1. The population of all fact types in <conceptual schema> is considered <closed_or_open> 2. The population of all fact types in <conceptual schema> is considered <state-of-affairs_or_actuality> 3. The population of <fact type> is considered <closed_or_open> 4. The population of <fact type> is considered <state-of-affairs_or_actuality> Note that a fact type overrides a conceptual schema specification. Note that there is a business rule that for each fact type it holds that it can have only one value of closed_or_open and one value of state-of-affairs_or_actuality.
The definition-based model specified in Clauses 8, 9, 11, 12 and 13 and the fact model defined in Clause 10 are different (although closely related) models. The differences between them should be described and a transformation from one to the other defined. This would address two concerns: 1. Separation of the two different meanings of 'fact type' into different models 2. Allow the definition-based model to have an open-world assumption and the fact model to have a closed-world assumption.
The overt problem is that SBVR has two different meanings for 'fact type': 1. In Clause 8, the extension of a fact type is currently a set of actualities, (although another issue proposes that this should be changed to a set of states of affairs) 2. In Clause 10, the extension of a fact type is a set of facts (propositions taken to be true). The underlying issue is: 1. SBVR's metamodel is defined in Clauses 8, 9, 11, 12 and 13. Its instances (domain models) are linguistic models of meanings. 2. The model defined in Clause 10 is included in the normative SBVR model to support a formal logic interpretation of SBVR's metamodel. Its instances (domain models) are fact models. The proposed resolution is: 1. State, in introductory text in Clauses 8 and 10, that the models are different 2. Somewhere in Clause10: a. List the major differences between the two models (see below) b. Describe informally the transformation to derive a domain fact model from a corresponding linguistic model. It is probably beyond the scope of this RTF to develop a formal specification of the transformation 3. Define fact models to be 'closed world' models. Another useful change would be to move the current Clause 10 so that it is placed after the clauses that define the SBVR metamodel - i.e. to renumber Clauses 11, 12 and 13 as 10, 11, 12 respectively, and renumber Clause 10 as 13. One of the reasons for raising this issue is the email discussion about Issue 14241 "Coexistence approach to SBVR" earlier this year. There, a case was made for allowing a fact model to be 'closed world', to enable it to be used as the basis for business applications that will run on relational databases using SQL. There was some discussion that SBVR was not primarily intended to model business applications; it was intended to model the business to be supported by these applications, and the models needed to be 'open world'. When the two kinds of model are recognized as being different, both needs can be satisfied: § The linguistic model defined in Clauses 8, 9, 11, 12 and 13 is the semantic community's open world model of its business. § The corresponding clause 10 fact model is the closed world model that is the basis for developing the data model and database for the required business applications. The transformations to these specifications are well-defined in both ORM and CogNIAM (the two fact modeling approaches described in Annexes of the SBVR Specification). One concern to be kept in mind is that the detail of specification in fact models (identifiers, data types, formats etc.) should not replicate capabilities already provided in other OMG specifications, especially UML and IMM. Dependencies with other Issue Resolutions Issue 10803: 'state of affairs' is an individual concept, not a thing If 'state of affairs' were deemed to be to be an individual concept, the argument here for the differences between the two models would not be substantially changed. Issue 13138: Move Fact Model Container Concepts from Clause 8 to Clause 10 A new issue has been submitted, proposing that the definitions in Subclause 8.5 be moved to Clause 13. Issue 14241: Coexistence approach to SBVR The proposed resolution here includes: 'closed world' assumption for Clause 10 fact model; fact type (Clause 8) having an extension that consists of states of affairs; fact type (Clause 10) having an extension that consists of facts. Issue (awaiting number): Each individual in the extension of a fact type (Clause 8) should be a 'state of affairs'. The proposed resolution here assumes that the extension of a fact type (Clause 8) is states of affairs. Resolution: See the discussion "Differences between Linguistic Model (Clauses 8, 9, 11, 12) and Fact Model (Clause 10)", below Revised Text: To be developed after discussion with RTF Disposition: Open
Subclause 8.5 is about the interchange files defined in Clause 15. The syntax for these files is (mostly) defined in Clause 13; the content of Subclause 8.5 should be placed in Clause 13.
From Issue 13138 (as of 5 Dec 2008): "Subclause 8.5 includes concepts conceptual schema and fact model that have no bearing on the content of the SBVR metamodel (as defined in the Clause 15.1 XMI file) or an SBVR model (to be illustrated by Clause 15.3 SBVR model of SBVR file). Rather they explain the structure of the SBVR model file in Clause 15.3 as an XML file containing a fact model population for an externally referenced SBVR XSD conceptual schema." The conceptual schema for interchange is the XSD, the facts are the XML content of the interchange file. Supporting arguments for making the change: § The specification does not place the syntax of Clauses 8, 9, 11 and 12 in Clause 8 - it is in Annex C § The specification does not place (most of) the syntax of Clause 15 in Clause 8 - it is in Clause 13 Some corrections are needed: § 'fact model' has two parts: 'conceptual schema' and 'fact population' § 'fact model is based on conceptual schema" should be 'fact population is based on conceptual schema' § 'conceptual schema includes fact' should be 'fact population includes fact' Dependencies with other Issue Resolutions Issue 13138: Move Fact Model Container Concepts from Clause 8 to Clause 10 This issue removes the definitions in Subclause 8.5 from the scope of Issue 13138. Resolution: Move the content of Subclause 8.5 into Clause 13, with the corrections listed in Discussion, above. Revised Text: TBD
'Actuality' is a specialization of 'state of affairs'. Clause 8 says: fact type (synonym: verb concept): concept that is the meaning of a verb phrase that involves one or more noun concepts and whose instances are all actualities There are other instances of fact type that need to be accommodated, such as: § states of affairs that are planned to become actualities § states of affairs that might be actualities, but the semantic community does not yet know for sure Instances of a fact type should be states of affairs.
See summary Dependencies with other Issue Resolutions Un-numbered Issue: Fact Type and Verb Concept should not be not synonyms This issue and the un-numbered issue "SBVR Linguistic Model and Fact Model are different models" together supersede "Fact Type and Verb Concept should not be not synonyms" Resolution: Change the definition of fact type to: fact type (synonym: verb concept): concept that is the meaning of a verb phrase that involves one or more noun concepts and whose instances are all states of affairs
The note under "state of affairs" reads: "A state of affairs can be possible or impossible. Some of the possible ones are actualities. A state of affairs is what is denoted by a proposition. A state of affairs either occurs or does not occur, whereas a proposition is either true or false. A state of affairs is not a meaning. It is a thing that exists and can be an instance of a concept, even if it does not happen. " Although unstyled, the use of "denoted by" is likely to confuse readers. The fact symbol "denotes" is used in clause 11.2.1.3 in the fact type "term denotes thing ". But a proposition is not a term, so this fact type is not what is meant in the note. The note is trying to use a passive version of "meaning corresponds to thing" from clause 8.6.1. Proposed resolution: 1. Add a synonymous form to "meaning corresponds to thing" such as "thing is meant by meaning". 2. Revise the note under "state of affairs" to use the new synonymous form and style the wording to make clear the reference to this formal SBVR concept.
Inconsistent use of terminology when relating facts to fact types It has been noted that there are a few places in clause 10 where the relationship between facts and fact types are described using inconsistent language. SBVR makes clear that not every fact is of a particular fact type – obviously, some facts are formulated using quantifiers, logical operators, etc. SBVR makes clear that instances of fact types are actualities, not facts. SBVR describes concepts as having instances, but not facts as having instances. A few places in clause 10 can be lead to confusion in this regard. They are listed below with recommended rewordings. Thanks go to Mark Linehan who graciously went through clause 10 last September and located these places. Recommended changes: 1. In the third paragraph of the introduction to clause 10, REMOVE the sentence that says: A ‘Fact’ is of a particular ‘Fact Type.’ 2. REPLACE the third paragraph of 10.1.1.2, which says this: The conceptual schema declares the fact types (kinds of facts, such as “Employee works for Department”) and rules relevant to the business domain. With this: The conceptual schema declares the fact types (such as “Employee works for Department”) and rules relevant to the business domain. 3. In the last paragraph of page 89 (in 10.1.1.2) there is a sentence that says: The fact model includes both the conceptual schema and the ground fact population (set of fact instances that instantiate the fact types in the schema). REPLACE it with this: The fact model includes both the conceptual schema and the ground fact population (set of facts that are formulated using the fact types and other concepts in the schema). 4. Just above figure 10.1 on page 90 there is the following sentence. Figure 10-1 provides a simplified picture of this situation, indicating that the fact model of sentences expressing population facts (instances of domain-specific fact types) is a varset (variable-set) whose population at any given time is a set of facts. REPLACE it with this: Figure 10-1 provides a simplified picture of this situation, indicating that the fact model of sentences expressing population facts (formulated using domain-specific fact types) is a varset (variable-set) whose population at any given time is a set of facts.
'Vocabulary' is defined in clause 11.1.3 as "set of designations and fact type forms primarily drawn from a single language to express concepts within a body of shared meanings ". 'Rulebook' is defined in clause 11.2.4 as "the set of representations determined by a given speech community to represent in its language all meanings in its body of shared meanings ". How does 'vocabulary' relate to 'rulebook'? When would an SBVR tool vendor use one or the other? The specification should either explain why it defines both these two concepts and when one would use one versus the other. --------------------------------
11.1.5.1 Kinds of Fact Type Problem Statement [Verb concept] templating could be interpreted to mean that SBVR gives templates *for* fact types, but that is not really the case. Template or 'templating' fails to accurately convey that the section is simply listing the common business-facing kinds of fact types practitioners would regularly want to define. Template or 'templating' connotes purpose, but a good name for a concept should indicate only essence. Proposed Resolution * A better signifier for the concept meant by verb concept templating should be based on the word structural. Structural is already accepted in SBVR for signifying things related to establishing the meanings of concepts (i.e., definitional matters). Specifically, it has been used in structural rules. * I used the term "element of structure" in Business Rule Concepts, 3rd Ed (several 1000 copies not distributed). So I would like to see some use of "structural" here. * Possible signifiers include "structural shape, "structural form", "structural purpose", "structural role" or "structural pattern". Note I the interest of moving forward with RTF work, I could live with synonyms for any use of "template" or "templating" in this section.
Describing the facts of a fact model, SBVR’s clause 10 says, “Population facts are restricted to elementary and existential facts.”
This “restriction” appears to be a restriction on the clause 10 mapping to a relational database, requiring a sort of normalization. It is certainly not a restriction discernable from SBVR’s definition of “fact model”. Nor is it a restriction on formal interpretation of fact models for knowledge bases in general. Facts that do not fall into those two categories (elementary and existential) can occur in fact models and can be mapped to formal logic. They can be formulated using concepts in a fact model’s conceptual schema, even if they cannot be formulated using those concepts in a way that is considered existential or elementary. Facts can be formulated using disjunction, universal quantification, etc.
A fact model can have a fact like the following, not as a rule in its schema, but simply as a fact:
“Every son of Mary has a car and a kayak”.
Whether this is a “good” fact in terms of being structured according to best practices is not relevant. Once we have a fact model, then we can use tools or guidelines to measure quality and recommend improvements. But that comes after we have fact model to examine.
Is the fact elementary? Not if it can break into “Every son of Mary has a car” and “Every son of Mary has a kayak”.
Is it existential? I cannot see it that way.
But it can map to formal logic, so clause 10 of SBVR should accommodate that mapping. It does not map directly into a relational table, but there is no reason to limit SBVR’s formal underpinnings to relational modeling.
As it turns out, clause 10 would handle the fact, “Every son of Mary has a car and a kayak”, just fine as long as it is formulated using a unary fact type as would be represented by a unary predicate like this: EverySonOfHasACarAndAKayak(Mary). That sort of contrived fact type is not likely to be found in a conceptual schema made up of meanings of words in a business vocabulary. Requiring a fact model with a business origin to have such a contrived fact type in its conceptual schema is inappropriate for SBVR, even though such contriving is sometimes part of database design. Conceptual schemas based on business vocabularies, rather than database design, involve meanings of words used by business people. Use of such vocabularies starts with an assumption that basic language works (quantifiers, conjunction, disjunction, restriction, demonstration, etc.) for putting words together to make statements. So formulations of facts so stated can tend towards complex formulations involving various sorts of quantifications, objectifications, logical operators, etc. Mapping such fact models into normalized databases is great, but requiring a direct mapping is not and must not be a limitation imposed by SBVR.
Some confusion is created in clause 10 from using the words “elementary” and “existential” as attributes of facts, when they seem to be attributes of formulations of facts, not of the facts themselves. For example, if the characteristic ‘employee number is assigned’ is define as “there exists an employee that has the employee number”, then by definitional substitution, these are two statements of the very same fact:
Employee number 777 is assigned.
There exists an employee that has the employee number 777.
So we have one fact that appears to be both elementary and existential. The difference is in formulation, not the fact.
It would be more clear for clause 10 to apply the ideas of “ground”, “elementary” and “existential” to formulation of facts rather than to facts. “Population” in the clause 10 sense seems to be strictly tied to formulation. It gives an example: “pop(Employee drives Car)= set of (employee, car) pairs …”.
Recommendation:
Remove the clause 10 general “restriction” to elementary and existential facts. Any such restriction should apply only to the clause’s relational mappings.
In clause 10, clarify how the concepts of “ground”, “elementary”, “existential” and “population” are tied to formulation.
The following wording was captured as part of the Issue 13716 notes, as part of some wording agreed in a long-ago meeting: From the meeting discussion notes on this Issue, the wording below was the agreed for the change instruction to Clause 8: This change has raised some concerns and, since it is not directly a part of the Resolution to Issue 13716, it needs to be its own issue.
In the course of discussion for Issue 13835 (re: "Fact Model"), I discovered what I believe to be a significant problem with the SBVR definition of "vocabulary" in Clause 11. To avoid complicating that original issue, I aim raising the problem here as a new issue. (Aside: I hope this new issue has not been overtaken by events ... it's been a long time since we've had a convenience document.)
Included in this document:
· pp. 1-2 Discussion and proposed resolution for the problem with "vocabulary" plus some additional observations about "terminological dictionary".
· pp. 3 (for convenience only) Mark's response (08:48 AM 6/28/2010) to my e-mail summarizing a resolution on issue 13835. Mark's response caused me to look closely at the SBVR definitions of "terminological dictionary" and "vocabulary".
· pp.4-7 (for convenience only) My original e-mail summarizing a resolution for issue 13835 (06/25/2010 07:54 PM).
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
DISCUSSION:
The current definition of "vocabulary" in SBVR reads as follows: set of designations and fact type forms primarily drawn from a single language to express concepts within a body of shared meanings
As far as I see, the definition says nothing directly or indirectly about *definitions*. This is inconsistent with (a) ISO, (b) MWUD, and (c) How real-world business people think of a "vocabulary". In these important ways, I believe the current SBVR definition is broken and needs to be fixed.
(a) ISO says (1087):
3.7.2 vocabulary
terminological dictionary (3.7.1) which contains designations (3.4.1) and definitions (3.3.1) from one or more specific subject fields (3.1.2)
NOTE The vocabulary may be monolingual, bilingual or multilingual.
RGR: Note the "and definitions (3.3.1)". We always based terms on ISO when we can - especially terms from their area of expertise.
(b) MWUD says:
1 : a list or collection of words or of words and phrases usually alphabetically arranged and explained or defined;
RGR: Note the "and explained or defined". This is the first and most common real-world meaning of "vocabulary".
(c) When business hear or say "vocabulary" they don't think simply of a list of words, they think of what the words *mean*. The words are of little use by themselves without definitions. Clause 11, the business-facing side of SBVR, *must* cater to commonly accepted usage of terms in the real-world.
PROPOSED RESOLUTION:
Change the definition of "vocabulary" in SBVR to be: set of designations and fact type forms primarily drawn from a single language to express concepts within a body of shared meanings and the definitions for those concepts
Also add: Source: based on ISO 1087-1 English (3.7.1) [vocabulary]
~~~~~~~~~~~~~
ADDITIONAL DISCUSSION
Re: "terminological dictionary"
Here is ISO's definition (expanded) ...
3.7.1 terminological dictionary
technical dictionary
collection of terminological entries (3.8.2) presenting information related to concepts (3.2.1) or designations (3.4.1) from one or more specific subject fields (3.1.2)
3.8.2 terminological entry
part of a terminological data collection (ISO 1087-2:2000, 2.21) which contains the terminological data (3.8.1) related to one concept (3.2.1)
NOTE Adapted from ISO 1087-2:2000.
3.8.1 terminological data
data related to concepts (3.2.1) or their designations (3.4.1)
NOTE The more common terminological data include entry term (3.8.4), definition (3.3.1), note (3.8.5), grammatical label (3.8.6), subject label (3.8.7), language identifier (3.8.8), country identifier (3.8.9) and source identifier (3.8.10).
RGR: The bottom line is that for ISO, "terminological dictionary" seems to be simply a more complete, formally organized version of a vocabulary. Both are listed along with other terms under the heading: 3.7 Terminological products.
I believe there is no reason not to stick as close as possible to the ISO sense of this term too(?). Otherwise, I question its usefulness for SBVR.
At 08:48 AM 6/28/2010, Mark H Linehan wrote:
Ron,
On the representation side, isn't "terminological dictionary" what you want? I note that "terminological dictionary expresses body of shared meanings " but from the Note under "terminological dictionary" it appears that should exclude the deontic rules. Perhaps if we define your concept "ABC" then we should say that "terminological dictionary expresses ABC ".
(On a related topic, I think that we should try to draw "conceptual schema" closer to "conceptual schema".)
--------------------------------
Mark H. Linehan
STSM, Model Driven Business Transformation
IBM Research
phone: (914) 784-7002 or IBM tieline 863-7002
internet: mlinehan@us.ibm.com
06/25/2010 07:54 PM
To: sbvr-rtf@omg.org
From: "Ronald G. Ross" <rross@BRSolutions.com>
Subject: [issue 13835 - "Fact Model"] Re: issue 13139 comments
All,
While memory is fresh, let me follow-up on yesterday's discussion in Minneapolis re the agenda item: Issue 13835 Use of the Signifier "Fact Model". I've now done some background research. Actually, there has been significant previous discussion of this topic, but largely under Issue 13139. (Unfortunately, I was unable to locate those e-mails in real time during the meeting itself.) The following analysis (organized into 10 key points) is longish, but aggregates everything into a single message for discussion and (my) future reference. Feedback welcome.
1. I believe Issue 13835 could really be called: Why won't "conceptual schema" or "fact model" as currently defined in SBVR work for Clause 11? To say it another way: What is the missing term for Clause 11?
2. This issue is a critical one. Like "rulebook", the missing term in Clause 11 represents a fundamental notion is *positioning* the purpose of SBVR from a business-facing point of view. Although not necessarily critical for software engineering(?), such positioning is absolutely central in establishing the appropriate market niche / mindset for SBVR itself.
3. It is increasingly clear that the missing concept should be one that *distinguishes* the business-facing side (and value-add purpose) of SBVR from the notion of "fact model" e.g., as in the ORM community.
4. To provide the widest possible umbrella as a standard, SBVR should accommodate that current understanding of "fact model" without change as much as humanly possible. (I believe it does.) SBVR should in no way 'step on' that pre-existing term. To do that was never our intention, of course, but we might have done that unknowingly.
5. Let's call the concept needed in Clause 11 'ABC'. What would an ABC look like? An ABC would ...
* have all the noun concepts and verb concepts (including individual concepts) you would need (to pre-define or adopt) in order to start in business tomorrow ("day one of business operations").
* thereafter, include any extensions to that necessary set of concepts based evolving business needs.
* primarily include elementary fact types.
An ABC would *not* include ...
* any deontic elements of guidance whatsoever.
* any ground facts you couldn't specify in advance of "day one of business operations".
6. The reason that "conceptual schema" doesn't work for ABC in Clause 11 is the following:
conceptual schema FL Definition: combination of concepts and facts (with semantic formulations that define them) of what is possible, necessary, permissible, and obligatory in each possible world
"Conceptual schema" includes deontic elements of guidance. It also treats what business people would call "rules" as "facts". That produces a completely unacceptable conflation of business-facing ideas. Business people simply don't say things like, "It's a fact there's a rule that ...".
7. The reason that "fact model" doesn't work for ABC in Clause 11 is the following:
fact model FL
Definition: combination of a conceptual schema and, for one possible world, a set of facts (defined by semantic formulations using only the concepts of the conceptual schema)
"Fact model" also includes deontic elements of guidance. It again treats what business people would call "rules" as "facts". (In addition, it would encompass ground facts created after "day one of business operations", as well as any and all non-elementary facts.)
8. The reasons that "body of shared meaning" and "body of shared meaning" don't work for ABC in Clause 11 are the following:
body of shared meanings
Definition: set of concepts and elements of guidance for which there is a shared understanding in a given semantic community
"Body of shared meanings" includes deontic elements of guidance. (In addition, it would encompass ground facts created after "day one of business operations", as well as any and all non-elementary facts.)
body of shared concepts
Definition: all of the concepts within a body of shared meanings
"Body of shared concepts" excludes all elements of guidance defined separately from definitions, including alethic ones. But definitional rules are most certainly involved in establishing a viable ABC. (In addition, it would encompass ground facts created after "day one of business operations", as well as any and all non-elementary facts.)
9. What can ABC be called? ISO has the term "concept system".
3.2.11 concept system
system of concepts
set of concepts (3.2.1) structured according to the relations among them
"Concept system" seems to be close to ABC. Since ISO did not consider rules, I think we can feel free to maintain that definitional rules would be covered by the definition.
Possible objections:
* The ISO definition doesn't seem friendly to unary fact types (" ... relations among them").
* It might be argued that the ISO definition would encompass ground facts created after "day one of business operations", as well as any and all non-elementary facts(?). (It's is not clear to me whether this is so.)
10. What does SBVR clause 11 really aim at? What signifier(s) best capture(a) the essence of ABC?
"verbal model" or "verbalization model" -
A major, indeed distinctive, goal of SBVR is to enable the expression of business rule statements and other forms of business communication is such manner that their full semantics can be captured and coordinated. We should emphasize SBVR's unique achievement in that regard by selecting an appropriate signifier, one that incidentally distinguishes ABC from "fact model" (and other 'structural' deliverables such as class diagrams and data models). For the past year or so, I have been using "verbal model" or "verbalization model" in my presentations for that purpose. They work well for that purpose.
MWUD
["verbal"]: 2 a : of or relating to words : consisting in or having to do with words
["verbalize"]: 2 : to state something in words : make a verbal statement
Note: If "concept system" is adopted from ISO, "verbal model" and/or "verbalization model" should be synonyms.
"structured business vocabulary" -
Clearly, that's what SBVR itself is -- SBVR has been described as a vocabulary for developing vocabularies. Like ISO (refer to the definition of "concept system"), we need to emphasize that Clause 11 is about creating a special kind of vocabulary, one that is *structured* (i.e., has verb concepts, etc.).
Note: "Structured business vocabulary" encompasses representation of meanings, not just meanings per se (i.e., it does not align with "concept system" in that regard).
The current definition of "vocabulary" in SBVR is:
vocabulary
Definition: set of designations and fact type forms primarily drawn from a single language to express concepts within a body of shared meanings
"Structured business vocabulary" can probably be defined as a synonym of "vocabulary". It's usefulness is that people don't normally think of fact type forms being in a vocabulary, yet that is a central, distinguishing characteristic of SBVR (i.e., to serve to support models for verbalization of business rules, etc.).
SBVR Clause 3 identifies ISO 6093 (Representation of numerical values in character strings) as a Normative Reference. SBVR 7.1.2 defines the symbol 'ISO 6093 Number Namespace' as a term for a namespace derived from a clause of ISO 6093. But there is no normative reference to the use of this namespace anywhere. Clause 8.7 says in a Note (informative) that ISO 6093 defines a set of designations for numbers, but it does not normatively specify that the ISO 6093 vocabulary is included in the SBVR Meaning and Representation Vocabulary. Either clause 7.1.2 or Clause 8.7 should say this normatively (if that is intended). Clause 13.2.7 refers to ISO 6093 in the (informative) Rationale section. Clause 13.2.7 defines the MOF representation of 'integer' to be the UML Primitive Type integer, but it uses CMOF:Class to represent 'number'. XMI 1.2 defines the exchange representation of CMOF:integer to be that defined for the "integer" type defined in XML Schema Part 2 Datatypes, and XML Schema Part 2 defines that representation directly without reference to ISO 6093. Nothing specifies the representation of instances of class "number". So, in terms of normative specification of signifiers for 'number', SBVR is not clear, and SBVR uses XML Schema Part 2 Datatypes, not ISO 6093, as the specification of signifiers for 'integer', which is said to be a specialization of 'number'. In practice, both standards specify the same representation for decimal numbers -- ISO 6093 NR2 and XML Schema 'decimal' -- but they state different rules for interpreting the precision of decimal fractions. The issue is completeness and consistency of the SBVR specification.
In SBVR clause 8.7, the terms 'quantity' and 'number' are marked as "FL", which means that they are formal logic concepts that are defined in Clause 10. The same is true of 'quantity equals quantity' and 'quantity is less than quantity'. Formal logic does not deal with physical quantities -- there is a whole science for that. And formal logic per se does not deal with numbers other than non-negative integers. The 'signed integer' concept is part of a specific mathematical theory. There is not, and should not be, any definition of these concepts in Clause 10. The FL marks should be removed. Further, these concepts should not be part of the Meaning and Representation Vocabulary, although they are useful business concepts that might be appropriate in Clause 11. Nonnegative integer is needed for the 'cardinality' concept; 'positive integer' is used in quantifications. 'Positive integer' is misused to represent an ordinal concept in 'starting character position' and as an identifier convention for instances of 'variable'.
Clause 8.7 introduces the idea of set and cardinality in order to support 'at least n' and 'at most n' constraint concepts. 'set' is defined to be an unordered collection of zero or more things. Marking 'set' a formal logic concept "FL" raises the issue of identity of things. Cardinality of a set is defined as "the number of distinct elements in the set, The definition of 'set' should also refer to 'distinct' or 'distinguished' things. The ability to distinguish makes it possible to determine the truth value of 'thing is in set' for an arbitrary thing. The 'set' entry should probably also include a Note, such as: Note: The means of distinguishing things as elements of a set is dependent on the kind of thing and the viewpoint taken in constructing each kind of set. Reference schemes may be used in this regard. Where the SBVR specification defines concepts that are 'sets', the defined reference scheme is used to distinguish elements.
In 11.2.1 we have an entry for something termed 'fact type role designation' -- its definition says that it is a "designation that represents a fact type role and that is not a placeholder " (See diagram, below.) There is nothing beyond a Definition for this concept. This entry doesn't make sense. I recommend it be dropped. (Or, if someone does see some usefulness for it, then please augment it with some notes and examples.)
The concept in SBVR Clause 8.1.1 defined as: “concept that is the meaning of a verb phrase that involves one or more noun concepts and whose instances are all actualities” has as its preferred term the signifier “fact type” This signifier, “fact type,” badly represents this concept and its definition. It is an example of bad term formation practice and is causing great confusion in the interpretation of the SBVR specification by contradicting the definition. Good term formation practice results in the best word or phrase that quickly and most reliably brings to mind the definition of the concept. In addition, this same signifier, “fact type,” is used as the term for a quite difference concept in Clause 10; thereby further increasing confusion in the SBVR specification. Recommended Resolution: Remove “fact type” as a term for the concept in SBVR Clause 8.1.1 that it currently represents, and replace it with the signifier “actuality type” as that is what the definition is defining.
In clause 8.3.4 of SBVR v1.0, the concepts: 'placeholder has starting character position' and 'placeholder uses designation' model the syntax of the non-normative Structured English language described in Annex C of the spec. These may not be properties of the syntax of other vocabulary and rules languages, and are unsuitable for graphical languages. The abstract syntax of any such language must be that a placeholder is an expression and must be unique within the fact type form. These requirements should be stated in the definition of placeholder. The placeholder expression is a designation for the role that is used only in definitions of the fact type, and its forms and roles. The idea of its "character position" is meaningless in graphical languages. The idea specified in 'placeholder uses designation' is a language convention that is not consistently used in SBVR and may well be different in other languages. The semantics of that syntactic construct is captured by 'role ranges over object type' in 8.1.1. Any convention for the syntax used by a tool is out of scope for SBVR. Therefore, both of these fact types should be deleted from the normative specification.
SBVR recognizes the notion of "property" in Clause 11.1.5 in "is-property-of", but never defines the concept directly. This omission should be corrected because "property" is a term used naturally by business people and business analysts. SBVR should own up to any term used commonly in the real world to form concepts and organize vocabulary. Resolution: Add the term "property" to Clause 11, defined as: Property: thing playing a role in a fact wherein the thing is perceived as being closely held by or descriptive of the thing playing the other role in the fact Dictionary Basis: a quality or trait belonging to a person or thing; [MWUD property] Necessity: The fact must be for a binary fact type. Example: In 'George was born on 22 February 1732', '22 Feb 1732' plays the *role* "birthdate", but "birth date" is a *property* of the *person* 'George'. The role has a *range* (date); the property has an *owner* (person). Example: "ceiling" denotes a property of a room and a property of an aircraft, two different properties of two distinct things
Problem: In clause 14.3, page 193, the example XML is wrong because it relates roles to the objectTypes ranged over using <sbvr:concept1SpecializesConcept2> instead of <sbvr:roleRangesOverObjectType> as required in the remainder of the specification, as shown in the diagram on page 192, and as shown in the "XML Patterns for Fact Types" in clause 13.6.4. I believe this is an editorial error that remains from when the SBVR FTF created the "role ranges over object type" verb concept. Also, the <sbvr:factType> element should be <sbvr:binaryFactType> and the <sbvr:designation> element should be <sbvr:factSymbol> On page 192, in the diagram, the box labelled ": fact type" should instead be labelled ": binary fact type", and the box labelled ": designation" (the one that is connected to the text box with "value=appoints") should instead be labelled ": fact symbol". Proposed Resolution: Update the diagram on page 192 as follows: - replace the text in the box labelled ": fact type" with the replacement text ": binary fact type: - replace the text in the box labelled ": designation" that is connected to the text box with "value=appoints", with the replacement text ": fact symbol" See this screen shot to identify the boxes that should be updated: Make these changes to the example XML on page 193: <sbvr:factType xmi:id="cao-c" role="cao-r1 cao-r2"/> --> <sbvr:binaryFactType xmi:id="cao-c" role="cao-r1 cao-r2"/> <sbvr:designation xmi:id="appoints" signifier="appoints-t" meaning="cao-c"/> --> <sbvr:factSymbol xmi:id="appoints" signifier="appoints-t" meaning="cao-c"/> <sbvr:concept1SpecializesConcept2 concept1="cao-r1" concept2="company-c"/> --> <sbvr:RangesOverObjectType role="cao-r1" objectType="company-c"/> <sbvr:concept1SpecializesConcept2 concept1="cao-r2" concept2="officer-c"/> --> <sbvr:RangesOverObjectType role="cao-r2" objectType="officer-c"/>
In clause 9.2.8, on page 71, the first example under "noun concept nominalization" is incomplete. The text says "In this example, ‘petrol’ is a mention of the concept ‘petrol’ which is used in the ‘type’ role of a fact type ‘quantity is of type’. " However, the formulation shown is missing the use of that fact type. Proposed resolution: Revise the example to read as follows. New/changed text indicated in red. Example: EU-Rent stores at least 300 kiloliters of petrol.” In this example, ‘petrol’ is a mention of the concept ‘petrol’ which is used in the ‘type’ role of a fact type ‘quantity is of type’. The statement is formulated by an at-least-n quantification. . The minimum cardinality of the quantification is 300. . The quantification introduces a first variable. . . The first variable ranges over the concept ‘kiloliter’. . The quantification scopes over an existential quantification. . . The existential quantification introduces a second variable. . . . The second variable ranges over the concept 'type' . . . The second variable is restricted by a noun concept nominalization. . . . . The noun concept nominalization binds to the second variable. . . . . The noun concept nominalization considers a projection. . . . . . The projection is on a third variable. . . . . . . The third variable ranges over the concept ‘petrol’. . . The existential quantification scopes over an atomic formulation. . . . The atomic formulation is based on the fact type ‘company stores thing’. . . . . The ‘company’ role is bound to the individual concept ‘EU-Rent’. . . . . The ‘thing’ role is bound to the first variable. . The at-least-n quantification is restricted by an atomic formulation. . . The atomic formulation is based on the fact type 'quantity is of type' . . . The 'quantity' role is bound to the first variable. . . . The 'type' role is bound to the second variable. (an alternate, and perhaps better, formulation would move the existential quantification of 'type' to the start)
Line 58 of the MeaningAndRepresentation-Model.xml file reads as follows:
<ownedMember xmi:type="cmof:Class" name="object type" xmi:id="objectType" superClass="concept"/>
The "superClass" attribute says that an Object Type is a kind of "Concept". This is inconsistent with clause 8.1.1, which defines 'Object Type' as a kind of 'Noun Concept'. This inconsistency causes problems (for example) when populating the "nounConcept=" attribute of the XMI tag <sbvr:closedProjectionDefinesNounConcept> because only a nounConcept can be referenced by this attribute, and an objectType is not a kind of NounConcept.
Proposed resolution:
Change line 58 of the MeaningAndRepresentation-Model.xml file to read:
<ownedMember xmi:type="cmof:Class" name="object type" xmi:id="objectType" superClass="nounConcept"/>
--------------------------------
Summary: There are two minor editorial issues regarding the verb concept "closed projection defines noun concept" in clause 9.3 1. In figure 9.12 on page 77 of the adopted specification and on page 79 of the ballot 3 convenience document, the verb concept is shown as "closed projection defines object type", rather than "... noun concept". Any noun concept should be definable this way, not just object types. The text is right and the graphic is wrong. 2. In the Acrobat Reader "Bookmarks" tab of the ballot 3 convenience document, the verb concept is shown as a sub-entry under "logical formulation constrains projection", rather than as a separate entry (as for "closed projection defines fact type". The problem occurs only in the convenience document, not in the formal adopted specification. See attached screen shot. Suggested Resolution: 1. Change the figure to match the text. 2. Fix the bookmark tab entry.
One of the example fact types provided in section 11.1.5.2 under “is-role-of fact type” is “rental car plays the role ‘replacement car’ in the fact type ‘breakdown during rental has replacement car’.” with the comment that “An instance of the fact type would be a particular breakdown during a particular rental having a particular replacement car.” I have a few concerns with this: 1. some of the text in this fact type should be in verb style 2. the underlining in ‘replacement car’ should be continuous both times 3. trying to instantiate the fact type produces something like “(The car registered) ’ABC123’ plays the role ‘replacement car’ in the fact type ‘breakdown during rental has replacement car’.” if we assume that underlined strings inside single quotes are not placeholders, while “(The car registered) ’ABC123’ plays the role ‘replacement car’ in the ??? ‘Breakdown #1234 has replacement car’.” is a more reasonable fact, except that a) this involves inconsistent handling of underlined strings inside single quotes, and b) ‘Breakdown #1234 has replacement car’ is neither a fact nor a fact type. 4. from this I deduce that the example seems to be a fact about the model rather than a fact type from which facts about EU-Rent can be generated 5. to support the latter argument, the EU-Rent examples in section E.1.4 has no ‘is-role-of’ fact types but does have ‘related facts’ such as “The noun concept 'return branch' is a role that ranges over the noun concept 'branch.’”.
The example fact type in section 11.1.5.1 under “is-property-of fact type” is “engine size is property of car model” yet the examples in Annex E do not have this form. Further if one tries to instantiate this fact type, one gets something like “351 cubic inches is property of Holden Marina” which misses essential information. I believe that ‘is-property-of’ fact types should each have the 2 forms “engine size of car model is cubic measurement”/“car model has engine size of cubic measurement” allowing for instantiations such as “engine size of Holden Marina is 351 cubic inches”/“Holden Marina has engine size of 351 cubic inches”.
Assortment fact types are not even fact types but facts since they make assertions about instances, “Graham Witt is a man” is of the same ilk as “Graham Witt is a citizen of Australia” (i.e. a fact).
The restriction of the definition of “business rule” to include only those rules that “the semantic community can opt to change or discard” is inappropriate. The SBVR definition of “rule statement” (“a guidance statement that expresses an operative business rule or a structural rule”) excludes those operative rules that are not business rules, for no obviously good reason.
“Each FemaleAustralian is a Person who was born in Country ‘Australia’ and has Gender ‘Female’” (section 10.1.1.2) and “Each Australian is a Person who was born in Country ‘AU’” (section 10.1.1.7) fly in the face of the meaning of citizenship: I was born in the UK but am an Australian, having taken out Australian citizenship, whereas Rupert Murdoch was born in Australia but is not an Australian, having renounced his Australian citizenship as a prerequisite of taking US citizenship. By the way these rules use a non-standard typography.
An elementary fact quoted in section 10.1.1.2 is “The Prime Minister named ‘John Howard’ was born in the Country named ‘Australia’” using yet another typography. This ‘fact’ is no longer true as, while John is still Australian-born he is no longer prime minister. An example, perhaps of the inadvisability of using role names in rules if they are not relevant to the rule. The following facts are correct but not time-dependent: a. The Man named ‘John Howard’ was born in the Country named ‘Australia’. b. The Man named ‘John Howard’ was Prime Minister of the Country named ‘Australia’ from 1996 to 2007.
In recognizing that an organization is not necessarily interested in recording all information about the real world, the SBVR proposes that there be two models of the world: a ‘reality model’ (of the real world) and an ‘in-practice model’ (of the organization’s view of the real world), which leads to some bizarre rule statements, listed below. Surely there is only 1 model in which both real-world objects and representations of them exist. The relevant quote from the SBVR is “Suppose the following two fact types are of interest: Employee was born on Date; Employee has Phone Number. In the real world, each employee is born, and may have more than one phone number. Hence the reality model includes the constraint ‘Each Employee was born on at least one Date’ (sic) and allows that ‘It is possible that the same Employee has more than one Phone Number.’ [If] the business decides to make it optional whether it knows an employee’s date of birth, [and] is interested in knowing at most one phone number for any given employee, … the in-practice model excludes the reality constraint ‘Each Employee was born on at least one Date’, but it includes the following constraint that does not apply in the reality model: ‘Each Employee has at most one Phone Number’. ” I believe there should be one model (not two), in which for each fact type there may be multiple rules reflecting specific requirements. Considering just dates of birth, the assertion “Each Employee was born on at least one Date” (which might be better worded as “Each Employee was born on exactly one Date”, “Each person has exactly one date of birth” or perhaps “Each person has a date of birth”) is a statement about the real world. Consider an insurance business that decides that it must collect the date of birth of each customer purchasing personal life insurance but does not need it for those purchasing only home insurance. Following the logic expressed in the SBVR (as quoted above) the ‘in-practice model(s)’ contain a new constraint: “Each person purchasing personal life insurance has a date of birth” (or “Each person purchasing personal life insurance must have a date of birth”) and an advice: “Each person purchasing only home insurance may not have a date of birth”. In fact the original assertion (“Each person has a date of birth”) still applies in the world view of the business, even to persons purchasing only home insurance. What is required is an additional constraint, which may be worded in one of the following forms “Each person who purchases personal life insurance must supply the date of birth of that person.” or “Each application for personal life insurance must specify the date of birth of the applicant.” and an advice “A person who purchases home insurance need not supply the date of birth of that person.”
In clause 9.2.8, in the entry for 'noun concept nominalization', there is an Example that begins:
"'EU-Rent stores at least 300 kiloliters of petrol.'
In this example, ‘petrol’ is a mention of the concept ‘petrol’ which is used in the ‘type’ role of a fact type ‘quantity is of type’.
The statement is formulated by an at-least-n quantification.
. The minimum cardinality of the quantification is 300."
This creates a dubious fact type and misconstrues "at least 300 kilolitres" as an at-least-n quantification.
"At least 300 kilolitres of petrol" is not an at-least-n quantification. It is not a reference to the cardinality of a set of distinct kilolitres that petrol has. (By way of analogy, my refrigerator stores about 3.5 litres of milk, which is clearly not a cardinality.) It is rather a comparison of two quantities -- the quantity (of petrol) stored and the quantity '300 kl' (of petrol). In SBVR SE, this statement should read:
"EU Rent stores a quantity of petrol that is greater than or equal to 300 kilolitres."
In a related previous issue, the FTF determined that a reference to "90 days" was an individual concept -- an amount of time. "300 kilolitres" is also an individual concept -- a 'quantity value'.
If the fact type in question is indeed 'company stores thing', then the 'thing' in question is an amount of a substance -- a 'quantity'. But 'quantity is of type' looks like a synonymous form of 'type has quantity', using 'of' as a verb, and that is altogether the wrong idea for the relationship. In fact, quantities are modifiers of nouns -- petrol (that is) in the amount of 300 kl -- but we don't need to introduce this complexity into the example.
In general, inventories are based on the fact type 'facility stores quantity of kind-of-thing. The point of the example -- that 'kind of thing' is a specialization of 'concept' and thus 'petrol' is mentioned/nominalized in this usage -- would not be marred by using this fact type and avoiding strange characterizations of quantities. Reformulating the example statement using this fact type emphasizes the noun concept nominalization and eliminates the confusing and erroneous elements of the example.
In SBVR C.1.6 there is an example, “thing [individual concept] is changed”, defined thus: “the extension of the individual concept is different at one point in time from what it is at a subsequent point in time”. In early SBVR thinking, the meaning of a singular definite description was an individual concept (a concept that corresponds to only one object [thing]) even if the description could refer to a different individual at a different time or in a different possible world. But that early understanding was later changed, as seen in a note in the SBVR entry for ‘individual concept’: “… each referring individual concept has exactly one and the same instance in all possible worlds”. Therefore, the first and third examples in C.1.6 and the similar example in E.2.3.1 need to be changed to not use ‘individual concept’. Perhaps a new concept type is needed for the meaning of a singular definite description.
All, In resolving Issue 15950 it has come to our attention that "community" and "semantic community" are used in Clause 12 in ways that are not really appropriate. I believe we are currently missing a very important concept for SBVR -- namely, the "business" part of "business rule". Attached is discussion and proposed resolution. Title: Governed Community & Adoption of Business Rules Source: Ronald G. Ross, Business Rule Solutions, LLC, rross@BRSolutions.com Summary: SBVR currently lacks a concept and term for the kind of community that creates business rules. This glaring omission was separated by agreement of the team from resolution of Issue 15959 (Inappropriate definitions of Business Rule, Rule Statement). The current definition of “community” is: group of people having a particular unifying characteristic in common The current definition of “semantic community” is: community whose unifying characteristic is a shared understanding (perception) of the things that they have to deal with By these definitions, any of the following could qualify as (semantic) communities: atheists, deists, communists, surfers, Francophiles, Anglophiles, futurists, business travelers, rappers, wine lovers, car surfers, baseball fans, diabetics, business travelers, psychics, nudists, philatelists, Egyptian protesters, Japanese earthquake victims ... Such communities do not, and cannot, create business rules. They lack the authority, standing and charter to do so. Unlike societies, organizations and businesses, they are not governed communities. Currently, SBVR has no concept for the special kind of communities that are governed. In effect, SBVR has no meaning for the “business” part of “business rule”. This omission is a significant one. In addition, SBVR currently does not adequately recognize or treat adoption of business rules. Adopting business rules is an act of free will (by a governed community) and should explicitly satisfy the “under business jurisdiction” test in the definition of “business rule”. Resolution: Add a category of “community” called “governed community” as follows. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Definition: community that by virtue of some recognized standing, authority or charter can create, adopt and apply business rules Dictionary Basis [MWUD “govern”]: 1a: to exercise arbitrarily or by established rules continuous sovereign authority over; especially : to control and direct the making and administration of policy in Examples: societies, chartered organizations, businesses, government bodies Example: EU-Rent is a legal entity, makes business rules for itself, and is therefore a governed community. Eu-Rent is also a member of each governed community (country) where it does business, as well as the European Union, a yet broader governed community. Note: A governed community can adopt sets of business rules (and advices) as-is, just like vocabulary. The decision to adopt business rules ‘as-is’ is an act of free will and therefore satisfies the “under business jurisdiction” test in the definition of “business rule”. Note: The “business” part of “business rule” is a popular, informal term for “governed community”. Note: The question “Who makes the rules?” for a governed community is outside the scope of SBVR. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Revised Text: Previously, I did a search of Clause 12, and sent my findings and recommendations. There are 5 segments of text where “semantic community”, “community” or “communities” appear. Below are (revised) recommendations for each. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ [1] body of shared meanings includes body of shared guidance Definition: the body of shared guidance is the set of all elements of guidance in the body of shared meanings uniting a semantic community that takes the elements of guidance as true RGR: This definition is problematic. Alethic elements of guidance might “unite” a semantic community (no real opinion), but I don’t see deontic elements of guidance as (a) “uniting” anything, or (b) pertaining to semantic community at all (unless the semantic community just happens to be a society, organization or business). Also, from a business perspective (as appropriate for Clause 11), a “community” doesn’t “take … elements of guidance to be true”. That’s a logician’s view. It would be more accurate to say ‘recognizes … as applicable’. Recommendation: Delete the phrase starting “uniting ...”. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ [2] business rule Definition: rule that is under business jurisdiction General Concept: rule, element of guidance Note: A rule’s being under business jurisdiction means that it is under the jurisdiction of the semantic community that it governs or guides - that the semantic community can opt to change or discard the rule. Laws of physics may be relevant to a company (or other semantic community); legislation and regulations may be imposed on it; external standards and best practices may be adopted. These things are not business rules from the company’s perspective, since it does not have the authority to change them. The company will decide how to react to laws and regulations, and will create business rules to ensure compliance with them. Similarly, it will create business rules to ensure that standards or best practices are implemented as intended. See subclause A.2.3. RGR: There are 3 instances of “semantic community” in this note. Recommendation: I would change this note to read as follows: Note: A rule’s being under business jurisdiction means that it is under the jurisdiction of the governed community that it governs or guides - that the governed community can opt to change or discard the rule. Laws of physics may be relevant to a governed community; legislation and regulations may be imposed on it; external standards and best practices may be relied upon. These things are not business rules from the company’s perspective, since it does not have the authority to change them. The company will decide how to react to laws and regulations, and will create or adopt business rules to ensure compliance with them. Similarly, it will create or adopt business rules to ensure that standards or best practices are implemented as intended. See subclause A.2.3. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ [3] advice of contingency Definition: advice of possibility that is a claim of contingency Note: The purpose of an advice of contingency is to preempt application of rules that might be assumed by some members of a semantic community, but are not actually definitional rules admitted by the community. Often, the reason for this assumption in a business is that other, similar businesses have such rules. Typically, the reason for providing such explicit advice is that people in the business have mistakenly applied the non-existent rule in the past. RGR: There is one instance of “semantic community” in this note and one instance of “community”. Recommendation: Both instances should be replaced by “governed community”. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ [4] advice of optionality Definition: advice of permission that is a claim of optionality Note: The purpose of an advice of optionality is to preempt application of rules that might be assumed by some members of a semantic community, but are not actually behavioral rules imposed by the community. Often, the reason for this assumption in a business is that other, similar businesses have such rules. Typically, the reason for such explicit advice is that people in the business have mistakenly applied the non-existent rule in the past. RGR: There is one instance of “semantic community” in this note and one instance of “community”. Recommendation: Both instances should be replaced by “governed community”. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ [5] Section 12.5, page 178, the paragraph that reads: In cases where definitions of concepts taken together do not logically imply something proposed in a structural rule statement, there is an inadequacy or mistake in either the relevant definitions or in the rule statement. The case of inadequate definitions is common and is acceptable in some communities. It occurs when a community shares a tacit understanding of many of its concepts. Words either have no explicit definitions or have definitions that use words that have no explicit definitions. Structural rule statements in this context can be correct, even if they logically follow from a tacit understanding of what characteristics are incorporated by concepts. RGR: There is one instance of “community” in this section and one instance of “communities”. Recommendation: I have no strong feelings at present about whether these instances should be changed or stand. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Clause 11 includes an entry for 'rulebook' (specifically, in 11.2.2.4). To maintain the separation of vocabulary-related items from rule/governance-related items (which has been the convention for Clauses 11 and 12), this should appear in Clause 12 rather than Clause 11. Resolution: Move 'rulebook' to Clause 12. [issue requested in the telcon of Mar. 18 2011]
Problem Description: The signifier "Explicitness of Representation" for a categorization scheme in SBVR 11.1.3 is not intuitive, and the reason for the choice is not explained. Explicitness of Representation Definition: the categorization scheme of the concept definition that classifies a definition based on whether it is owned by its speech community or adopted by its speech community Resolution: Change the signifier for the concept to "Origin".
Issue Title: Conflation of the signifier “rulebook” with the concept/definition for Speech Community Representations (a container concept /set) Clause: 11.2.2.3 Printer Page: 155 Issue Statement: The concept (definition) in Clause 11.2.2.3 defined as: the set of representations determined by a given speech community to represent in its language all meanings in its body of shared meanings is conflated with the undefined concept most commonly associated with the signifier “rulebook.” The set defined in this entry is only the representations for one speech community and does not include any semantic connections between meanings, which are required to compose the content of a rulebook. Proposed Solution: Separate the two concepts by creating a new entry for “rulebook”; provide a definition for rulebook that can be used to produce one; and replace the signifier “rulebook” on the existing entry with “speech community representations”.
SBVR is not clear about how markup should or should not be embedded within Representation Expressions. The specification needs to be clear about exactly what is included in basic Representation Expressions, especially Fact Type Forms, which contain no embedded markup. It also needs to be clear about the kinds of markup that can be embedded in Representation Expressions and how to communicate which markup specification is being used.
There appears to be something missing ("is" -- or, the more verbose, "that is") in the Definition given for "expression" (p. 26 -- PDF p. 38),
i.e., ..."but is independent"... (... "but that is independent"...).
Summary: SBVR makes an important distinction between the meanings of the word “exists” (existential quantification) and the word “occurs” (used to describe a state of affairs). A state of affairs can exist and thereby be involved in other things (e.g., plans, desires, fears, expectations) even if it does not occur, even if it never occurs. SBVR should explicitly define and explain the characteristic ‘state of affairs occurs’, and should then use that characteristic to define ‘actuality’. Note that this issue is related to issue 14849 and became important in discussing 14849, but its resolution should be independent of 14849.
In clause 8.3.3, in the glossary entry for "statement", SBVR has the Necessity "Each statement expresses exactly one proposition ". This Necessity is also shown in figure 8.4 and is cited as an example on printed page 6. The issue is that some statements do not express propositions (i.e. a meaning that is true or false, per the definition of 'proposition' in 8.1.2). There are at least two types of statements that are neither true nor false: (a) paradoxes, such as "This statement is false"; (b) atemporal statements used with temporal worlds. For example, the statement "the board of director meets" is a proposition (i.e. either true or false) in an atemporal world (i.e.a world that only contains facts about one moment in time). But in a world that has records of multiple meetings of the board of directors, the statement is ambiguous. It can be understood as true if read as meaning "the board of directors meets at some time". It is either true or false (according to the facts in the world) if it is read as "the board of directors meets right now". Clearly a statement does not express a proposition when the statement is paradoxical or ambiguous. Suggested resolution: Revise the Necessity to read "Each statement expresses at most one proposition." Revise the figure and the example to match
Clarify that objectifications based on a fact type can refer not only to actualities, but more generally to states of affairs, regardless of whether they are actual. Fix examples of objectifications to include objectifications of states of affairs that are not necessarily actual. Also, for SBVR Structured English in the explanation of using the demonstrative “that” for objectification, refer more generally to “state of affairs” rather than to “actuality”.
SBVR defines the concept "thing" in clause 8.7. The definition is unclear as to whether the extension of "thing" contains only singletons (i.e. individual things) or can contain instances that recur in some way. Proposed Resolution: Add a Necessity or Possibility or Note that explains whether individual things can recur. Add examples.
In recent RTF teleconferences, it was agreed that in Clause 11.1.3, Kinds of Definition, some additional notes are needed for “adopted definition” to explain that adoption of a definition is the mechanism for adopting the meaning of a concept.
The mechanism for adopting (the meaning of) a concept is: • A meaning cannot be adopted in the abstract. A representation of the meaning – a definition - is what is adopted. • A definition is expressed in some language, so is adopted by some speech community within the adopting semantic community. • The first definition adopted by a semantic community (via one of its speech communities) is the adoption of the concept • Subsequent definitions of the adopted concept (e.g. in other natural languages) must have the same meaning as the first adopted definition When a definition is adopted, the adopting speech community may use a term or verb symbol different from the one in the source. For example, SBVR has adopted the definition of ‘object’ from ISO 1087, but uses the term ‘thing’ to designate it. Related adopted definitions may be localized while retaining their meanings. For example, SBVR has adopted the definition of ‘individual concept’ (‘concept that corresponds to only one object’) from ISO 1087, but has localized it as ‘concept that corresponds to only one thing’. Dependencies with other Issue Resolutions 16059 Governed Community & Adoption of Business Rules The mechanism for adopting concepts described here is the basis for adopting business rules as described in the resolution of Issue 16059.
SBVR’s explanation of the concept ‘state of affairs’ could be improved by clarifying how states of affairs include or exclude each other. This is relevant to distinguishing involvement (already defined in SBVR) from inclusion. It is also relevant to understanding the relationship between a situation and the circumstances it includes
"Objectification" is currently defined in Clause 9.2.7 to be a logical formulation. This usage of the term is counterintuitive for several reasons. (1) Logical formulations are a way of structuring meaning particular to SBVR. (2) Objectification can be used to mean the 'process' of objectifying, rather than the *result* of objectifying, as usually preferred in SBVR.
Resolution:
1. Change each instance of "objectification" in Clause 9.2.7 to "objectifying formulae".
2. Inspect every other instance of "objectification" in SBVR to determine whether it refers to "an objectifying formulae" or to the process of objectification ("objectifying"), and adjust accordingly.
3. Add concepts, definitions, and terms for the three kinds of *results* from the process of objectification, and if appropriate, a more general concept for the three (probably called objectification).
Note: At least one of these three kinds of objectification, the one pertaining to open variables, should be included Clause 11.1.5. Probably all three should be. "Objectification" (meaning the result of objectifying) is clearly an 'element of structure' in the sense of 'characterization', 'categorization', etc. (albeit more complex).
Problem Statement: "Nominalization" is currently defined in Clause 9.2.8 to be a logical formulation. This usage of the term is counterintuitive for several reasons.
(1) Logical formulations are a way of structuring meaning particular to SBVR.
(2) "Nominalization" can be used to mean the 'process' of nominalizing, rather than the *result* of nomalization, as usually preferred in SBVR.
(3) "Nominalization" should be included in SBVR under its real-world (MWUD) meaning.
Resolution:
1. Change each instance of "nominalization" in Clause 9.2.7 and 9.2.8 to "nominalizing formulae".
2. Inspect every other instance of "nominalization" in SBVR to determine whether it refers to "a nominalizing formulae" or to the process of nominalization ("nominalizing"), and adjust accordingly.
***Note: This includes the definition of the critical term "state of affairs" (in the convenience document available as of 8/2011).
3. Add concepts, definitions, and terms for the three kinds of *results* from the process of nominalization, and if appropriate, a more general concept for the three (probably called nominalization).
Note: It needs to be determined where in SBVR these entries should be included.
Problem Statement: 1. "Aggregation" is currently used on page 47, but as far as I can tell is not defined anywhere. "Aggregation" should be included in SBVR under its real-world (MWUD). meaning. What is it meant to be (real-world sense). 2. For consistency, "Aggregation Formulation" should probably be renamed "Aggregating Formulae". See other issues submitted concerning "objectification" and "nominalization". Resolution: 1. "Aggregation" should be included in SBVR under its real-world (MWUD) meaning, and included in the appropriate section. 2. Change each instance of "aggregation formulation" in "aggregating formulae".
Problem Statement: "Projection" is currently defined in Clause 9.3 to be a semantic formulation. This usage of the term is counterintuitive for several reasons.
(1) Semantic formulations are a way of structuring meaning particular to SBVR.
(2) "Projection" can be used to mean the 'process' of projecting, rather than the *result* of projection, as usually preferred in SBVR.
(3) "Projection" should be included in SBVR under its appropriate real-world meaning.
Resolution:
1. Change each instance of "projection" in Clause 9.3 to "projecting formulae"
2. Inspect every other instance of "projection" in SBVR to determine whether it refers to "a projecting formulae" or to the process of projection ("projecting"), and adjust accordingly.
3. Add a real-world concept and definition for "projection" and for "bag" as currently used in "bag projection". (It needs to be determined where in SBVR this entry should be included.)Problem Statement: "Quantification" is currently defined in Clause 9.2.6 to be a logical formulation. This usage of the term is counterintuitive for several reasons.
(1) Logical formulations are a way of structuring meaning particular to SBVR.
(2) "Quantification" can be used to mean the 'process' of projecting, rather than the *result* of projection, as usually preferred in SBVR.
(3) "Quantification" should be included in SBVR under its appropriate real-world meaning.
Resolution:
1. Change each instance of "quantification"" in Clause 9.3 and elsewhere to "quantifying formulae"
2. Inspect every other instance of "quantification" in SBVR to determine whether it refers to "a quantifying formulae" or to the process of quantification ("quantifying ), and adjust accordingly.
3. Add a real-world concept definition for "quantification". (It needs to be determined where in SBVR this entry should be included.)
SBVR clause 8.1.2 defines 'proposition' as 'meaning that is true or false'. The Date/Time specification, and some SBVR examples, show that some propositions are used for their "content" -- the situation that the proposition describes -- without regard to their truth value. For example, "Each rental car must be inspected before it is available for rental" uses the proposition 'rental car r is inspected' (for each referent of r) to refer to situation in which the car is inspected, and the proposition 'rental car r is available for rental' to refer to the situation in which the car can be rented. The rule relates these situations without requiring any true/false evaluation of either of them. Further, the situation in which a given rental car is available is only sometimes an actuality; the proposition 'r is available for rental' can be sometimes true and sometimes false in the actual world. Thus, being true or false is not the most important characteristic of a proposition, and may not be well-defined. Recommendation: 'proposition' should be defined as: conceptualization of an event, activity, situation or circumstance. Such a definition would be consistent with the idea that it 'corresponds to' a 'state of affairs'. It is also consistent with the idea that true and false are defined in terms of correspondence to an actuality. Those properties would be dependent on the situation that is identified in the proposed definition. This change of definition does not change the intent of the term 'proposition' in any way. It just avoids having the concept depend on having a truth value in usages that don't care. (It may be that the proposed definition needs some additional characteristic to distinguish it from a noun concept that corresponds to events, like 'heart attack'. For example, the proposition must be based on one or more fact types and involve things in fact type roles.)
Definite descriptions do not always define individual concepts
The entry for ‘definite description’ in SBVR 11.1.3 includes this structural rule:
Necessity: Each definite description is the definition of an individual concept.
The rule is incorrect. A definite description defining a concept in a schema might well be taken as defining an individual concept, but a definite description within a statement of a fact in a model need not define an individual concept because it need not identify the same individual in all possible worlds. It would identify an individual in the world described by the fact. Similarly, a definite description in the context of a rule statement might identify a single individual in each situation addressed by the rule, but not necessarily the same individual in all possible worlds. E.g., “the previous calendar month” definitely describes one month, but which month it describes depends on the current month, which can vary across possible worlds.
Also, a note should be added to the entry for “definite description” to point out that the one thing defined by a definite description can be a set (e.g., “the cars owned by EU-Rent”, which, by the way, is not the same set in all possible worlds).
Is a 'boolean variable' a proposition? It is defined to be a variable whose referents are truth values, and I have no idea whether it is a 'meaning'. I believe 'variable' is used in SBVR in the sense of 'formulaic variable' ... but it's not clear from its definition alone. The point needs to be clarified; otherwise, it will only continue to cause problems. We shouldn't have real-world words being used in a special sense in SBVR
Disposition: Resolved OMG Issue No: ???? Title: Need business-oriented verb concepts to support "local closure" Source: Mark H. Linehan, IBM Research Summary: Clause 10.1.1.3 has an extensive discussion of "Open/Closed World Semantics". In particular, the penultimate paragraph near the bottom of page 94 of version 1.0 of the specification says: "For any given schema, the business might have complete knowledge about some parts and incomplete knowledge about other parts. So in practice, a mixture of open and closed world assumptions may apply. We use the term “local closure” (or “relative closure”) for the application of the closed world assumption to just some parts of the overall schema. One might assume open world semantics by default, and then apply local closure to specific parts as desired; or alternatively, assume closed world semantics by default and then apply “local openness.” We adopt the former approach as it seems more realistic when modeling real business domains." In SBVR 1.0, local closure is supported by the verb concepts "fact type is internally closed in conceptual schema" and "concept is closed in conceptual schema" in clause 8.5. The resolution of issue 13138 moves clause 8.5 to clause 10, thus making these verb concepts no longer available in the normative specification or in the clause 15 supporting documents. The result is that the specification no longer supports the semantics mentioned in the quote given above. This issue requests that similar functionality be added to clause 11. The original clause 8.5 verb concepts used designations that are not meaningful to business people. The resolution of this issue should adopt business-oriented terminology. Discussions have identified at least four possible approaches: 1. A verb concept "set is completely known", meaning that the semantic community knows all the elements of the set. This would be particularly useful when applied to a set as the extension of a concept. 2. A verb concept "concept has completely known extension". Similar to the above, but applying specifically to the extension of concepts. 3. A verb concept such as "semantic community completely knows concept". 4. Building on the concept "communication concept" in clause 11.2.2.3 to define closure with respect to an information record. Example use cases for local closure include the following: Example 1 This example is about a concept called order that includes a list of line items, where each line item has a quantity, a catalog id, etc. A minimal vocabulary is shown here, just enough to illustrate the example. order Definition: A customer request for one or more products and a promise to pay the total cost of the order. line item Definition: Details about an order for a particular product. quantity Definition: positive integer that is the number of units of the product that is desired by the customer catalog id Definition: text that identifies the product desired by the customer line item has quantity Necessity: Each line item has exactly one quantity. line item has catalog id Necessity: Each line item has exactly one catalog id. order includes line item Necessity: Each order includes at least one line item. "order includes line item" is internally closed in the business xx conceptual schema The "internally closed" fact says that the business knows all the line items that are included in each order: there are no other line items. Consider a rule such as "Each order must be shipped within 24 hours if the order does not include a line item that has quantity greater than 100." As described in clause 10.1.1.3, this rule makes no sense with the default SBVR "open world" semantics because under those semantics, the business cannot know that no "line item that has quantity greater than 100". Example 2 Consider a business that has a vocabulary about employees. The business considers it knows all its employees; there are no employees that it does not know. employee Definition: person that works for the business Under SBVR's default open world semantics, the glossary entry given above is insufficient because it does not capture the business sense that it knows all its employees. To accomplish that, the vocabulary needs the following: "employee" is closed in the business xx conceptual schema Example 3 Continuing example 2, suppose the business needs concepts relating to employee names and work phone numbers: employee name Definition: text that identifies an employee work phone number Definition: number used to phone an employee at work The business requires that it knows the employee name of each employee because the government requires this information on tax and employment reports. So the employee name is authoritative. The business knows that, in practice, it does not know the work phone number of each employee. These change too often to keep up with. SBVR needs verb concepts to express the idea that the employee name is reliably know, but the work phone number is not reliably known. Resolution: Revised Text: Disposition:
SBVR says (in clause 8.6.2, as of ballot RTF 1 ballot 5) that "Each proposition corresponds to exactly one state of affairs." For example, the proposition "each driver of a rental is qualified" (as may be embedded within an obligation statement) corresponds to a single state of affairs in which all drivers of a rental are qualified. Per clause 8.1.2, such a proposition is true or is false according to whether the corresponding state of affairs is actual. This idea is meaningful to logicians but not to business people. Business users of SBVR will not care about a state of affairs in which "all drivers of a rental are qualified". What is meaningful to business users is the actualities that comprise that state of affairs – in this case, whether each driver, taken individually, is qualified. If the overall proposition is false, an immediate question will be, "which driver is not qualified, and why not?" To support this kind of analysis, SBVR should have a verb concept that relates a proposition to the actualities that make the proposition true or false. The relationship already exists indirectly through the "state of affairs1 includes state of affairs2" verb concept introduced by the disposition of issue 16526. The current issue proposes a direct relationship, built on and consistent with "state of affairs1 includes state of affairs2", that avoids the need for business users to understand the logician's idea of "proposition corresponds to state of affairs".
Clause 10 of SBVR provides a formal logic interpretation of SBVR in terms of Object Role Modeling (ORM). There has been a long-standing agreement within the OMG community to provide a formal interpretation in terms of Common Logic (CL). CL is an ISO standard (ISO 24707) for which there is an OMG standard metamodel in the Ontology Definition Metamodel (ODM) specification, and which is being used as a basis for logical interpretation in the OMG Date Time vocabulary. A partial interpretation of SBVR in CL is given in clause 10.2, but significant work is needed to complete this grounding. Completion is essential to supporting downstream alignment of OMG specifications that are expressed in terms of other logic languages, to reuse of SBVR vocabularies by commercial rule engines, and to facilitate interoperability with other work in the ISO community. It may also be needed to support development of new vocabularies in SBVR, such as potential financial services vocabularies related to the FIBO (Financial Industry Business Ontology) effort in the Finance DTF.
Spin-off from Issue 14843 (via Issue 15623 Issue Resolution into which it was Merged) The definition-based model specified in Clauses 8, 9, 10, 12 and 13 and the fact model defined in Clause 10 are different (although closely related) models. The differences between them should be described and a transformation from one to the other defined. This would address two concerns: 1. Allow the definition-based model to have an open-world assumption and the fact model to have a closed-world assumption. The proposed resolution is: 1. Define that Clause 10 ‘fact models’ are by default closed world models
Spin-off from Issue 14843 (via Issue 15623 Issue Resolution into which it was Merged) The definition-based model specified in Clauses 8, 9, 10, 12 and 13 and the fact model defined in Clause 10 are different (although closely related) models. The differences between them should be described and a transformation from one to the other defined. The underlying issue is: 1. SBVR’s metamodel is defined in Clauses 8, 9, 10, 12 and 13. Its instances (domain models) are linguistic models of meanings. 2. The model defined in Clause 10 is included in the normative SBVR model to support a formal logic interpretation of SBVR’s metamodel. Its instances (domain models) are fact models. The proposed resolution is: 1. State, in introductory text in Clauses 8 and 10, that the models are different 2. Somewhere in Clause 10: a. List the major differences between the two models b. Describe informally what transformation would be needed to derive a domain fact model from a corresponding linguistic model. It is probably beyond the scope of this RTF to develop a formal specification Resolution: 1. Add a subclause to Subclause 10.1.1 to discuss to an appropriate level of detail all aspects of the relationship between the concepts in the SBVR Vocabularies in Clauses 7, 8, 9, 11 & 12 and the formal interpretation in Subclause 10.1.1, as well as removing ambiguity from Clause 10.1.1 by consistent use of terms intension, extension, fact population, and the set of all possible facts..
OMG Issue No: 16685 Title: Fix Entries in Subclause 10.1.2.1 to Align with Subclause 10.1 Source: SBVR Co-chair, Donald Chapin [Donald.Chapin@BusinessSemantics.com] Summary: Spin-off from Resolution of Issue 15623 (and 14843 which was Merged into it) Fix the entries in SBVR Subclause 10.1.2.1 to bring them in line with what Clause 10.1 says as revised by the resolution to Issues 15623 & 14843.
(a) Clause 11 should include the verb concept "thing has property". This verb concept should appear in figure 11.5. (b) Property needs to be indicated as an abstract concept in Clause 13 (since it is in the universe of discourse, not the model).
Section 13.2.5 is based on a misunderstanding and misuse of MOF: a) the phrase “In each case where an attribute and an association end represent the same role, the SBVR Metamodel includes a tag that tags both the attribute and the association end. “ is in ignorance of the fact that in MOF2 and UML 2 that “attributes” and “association ends” are both represented as instances of Property: and in such a case there would be a single instance of Property linked to the Class using the attribute meta-property and from the Association by the memberEnd (or ownedEnd) property: eliminating the need to link two separate elements. b) attempting to use MOF Tags to link two properties. In fact MOF Tags are “Simple string name-value pairs”. This is an urgent issue since it affects the production of the SBVR 1.1 artifacts: there is in fact no need for the tags that have caused some of the difficulties producing the machine-readable files: even the file I sent today for the metamodel, which uses the Tag value property does not match this section of the spec which states that value is the empty string.
SBVR makes use of ElementImports to give additional aliases to some elements in the same package. This is invalid use of ElementImport: the UML/MOF specs clearly state that ElementImports are only for elements “in another package”. I recently confirmed my understanding with the UML team that “another” does mean literally that (confirmed by OCL elsewhere in the spec) and it cannot be interpreted to mean “the same package”. Even were the ElementImport to be permitted, it would not have the intended meaning which I believe is to add additional synonyms to elements. In contrast the alias in an ElementImport “Specifies the name that should be added to the namespace of the importing Package *in lieu of* the name of the imported PackagableElement.” This issue is urgent since it affects the production of correct normative artifacts for SBVR 1.1.
The value of a comprehensive and coherent SBVR example seems to be generally accepted, but there have been some concerns expressed about the size and complexity of the EU-Rent example (Annex E). Section E.2.2.1.1 Car Movement is a particular problem. It is presented first in the detail of EU-Rent‘s vocabulary but is quite complex. It introduces the idea of ‘car movement’, a component that is used in two different contexts as part of the definition of a rental, and as part of the definition of a logistical car movement made by a EU-Rent employee. Annex E could be made more digestible, without substantially changing its content, by: 1) Presenting the sections in a different sequence, with sections that introduce simpler ideas presented earlier. 2) Presenting ‘car movement’ in a simpler form This issue can be resolved alongside Issue 10628: Align Annex E with the normative text. To avoid delay in updating the SBVR specification, updating EU-Rent to comply with the SBVR Date-Time Vocabulary is outside the scope of this issue, and will be addressed later.
Problem: SBVR adheres to ISO 1087 as much as possible. Clearly evident in the structure of ISO 1087 is that no definition should appear until all terms within it have been defined (as needed). This best practice (rule) results in a logical, easy-to-follow presentation. Clause 11 of SBVR is clearly broken in this regard. The result is significant lack of clarity, making detection of errors unnecessarily difficult. The possibility of misinterpretation (or non-comprehension) by software engineers and other readers is high. Aside: Personally I think this solution amounts to simple editing because anyone could apply the ISO 1087 rule without understanding a thing about the content. However, since some might see new headings or groupings as somehow conveying meaning – never the case in SBVR – I have nonetheless requested an issue. There are also a few choices about optimizing placement. Note: **This issue can be resolved without using any meeting time.** Resolution: Apply the ISO 1087 rule about sequencing vocabulary entries rigorously. This re-sequencing requires no changes in the entries themselves. Attachments: Two files containing all Clause 11 entries are attached. One file is unchanged except that entries are numbered. The other file is re-sequenced. The entry numbers reappear in the re-sequenced file indicating the original location of each entry. Unfortunately, this working version does not use the latest version of SBVR … I did not have the source file for that. However, since no changes to the entries themselves are covered by this issue, the version used is largely immaterial to illustrate the proposed resolution. The lay-out simply needs to be re-done for the newer material.
Problem: The Revised Text of 17017 makes no mention of clause 13.2.2. In clause 13.2.2, the first paragraph contains the sentence: "The signifier of each synonym of the designation is an alias for the class." Nothing is said in 13.2.2 about how to encode the "alias", but the diagram in 13.2.2 shows an "element import". The revised text does not, but should, delete this drawing element as well. Further, under the Rationale subhead in 13.2.2, the first sentence reads: "Use of aliasing, though not common in MOF-based metamodels, keeps a strong alignment of the SBVR Metamodel with the SBVR vocabulary." Presumably, that will no longer be the case if the element imports are deleted. I suggest it should rather read: "In general, MOF does not provide a mechanism for declaring synonyms. Therefore, the Synonym elements of the SBVR vocabularies do not have counterparts in the SBVR MOF metamodel. They are, however, captured in SBVR vocabularies that are instances of the SBVR MOF metamodel."
"vocabularies" is miss-spelled "vocabulaires" in the sixth paragraph of clause 10.1.1 in convenience document 8.
Annex H provides detailed guidance on the representation of SBVR vocabulary concepts in UML diagrams. Much of that guidance produces invalid UML constructs per UML 2.4. H.1 "If there are additional terms for the concept they can be added within the rectangle, labeled as such -- e.g., “also: is-category-of fact type” as depicted in Figure H.1." There is no UML syntax for this. H.2 "Alternatively, an individual concept can be depicted as an instance of its related general concept (noun concept), as in Figure H.3." The diagram uses an unidentified Dependency, which has no meaning. It should be formally stereotyped. H.3.1 shows three representations of the fact type 'semantic community shares understanding of concept'. The third is invalid -- an association can have only one name. Also the name of the association is 'shares understanding of'; it does not include the placeholder terms. H.3.1 Figure H.4 shows associations that are navigable in both directions, inducing unnamed UML properties on 'semantic community' and 'concept' that are not intended. (This is a vestige of UML v1 ambiguity.) It should show no navigable ends, using UML 2.4 syntax. H.3.4 Figure H.9 depicts an invalid relationship symbol; an association is required to have 2 or more roles. H.4.2 Figure H.11 shows a stereotype <<is role of>> on a Generalization. I'm not sure this is valid UML, but in any case such a stereotype would have to be defined in a formal Profile. (Semantically, some "roles" are object types that specialize more general concepts, others are association ends (verb concept roles), and others are things in their own right that have the property 'role has occupant'.) H.4.3 suggests that there is no consistent mapping for association names. In any case, the UML model of a 'fact type role' is a named association end, regardless of ownership. H.6.1 Figure H.14. It is not clear what UML element has the name "Results by Payment type", and the text does not say. It may be a GeneralizationSet. H.6.2 Figure H.16. ":modality" appears to be a TagValue associated with some unnamed and undefined Tag, or it may just be another string that names no model element. H.8 In, Figure H.17 there is a meaningless dashed line between 'car recovery' and a ternary association (verb concept). It is said to represent 'objectification'. That dashed line should be a Dependency that has a stereotype indicating the nature of that relationship, e.g., <<objectification>>, defined in a Profile. H.9 says that the default multiplicity on association ends is 0..*. According to the UML metamodel v2.4, the default multiplicity on a UML association end is 1..1, i.e., exactly one. This makes most of the SBVR UML diagrams implicitly erroneous. So Annex H needs to be rewritten, and if it is to include standard stereotypes and tag values, it needs a standard UML Profile that defines them. Further, it demonstrates the need for minor repairs to the UML diagrams throughout SBVR, to make them match the MOF model described in Clause 13.
The precedence of logical operators ("and", "or", etc.) in Structure English is unspecified which may make some rules ambiguous. Furthermore, they sould be called "operators" and not "operations".The Structured English keyword "another" is sometimes ambiguous. For an example, in the following rule, it is formally not clear whether "another <person3>" refers to <person1> and/or <person2>: It is prohibited that a <person1> <is married to> <person2>, if that <person1> <is married to> another <person3>.
SBVR apparently assume that business terms are composed of natural language words, and that those words may have multiple morphemes that are nonetheless the same word and the same term. That is, a vocabulary term like 'purchase contract' may also have the form 'purchase contracts', and a vocabulary term like 'is owned by' may be expressed as 'has been owned by'. But SBVR says nothing about any of this in defining 'designation' or 'signifier'. When a signifier for the same concept is in a formal language like OWL or CLIF, this idea of multiple morphemes is not (usually) part of the language syntax. So this should be carefully addressed. For the SBVR Structured English language, Annex C.1 explicitly says that these alternate morphemes are "implicitly available for use", which may mean they are treated as equivalent, or just that they are recognized as uses of the same designation. In natural language, such morphemes carry additional meaning , e.g., plurality or tense or mood. And a morphological variant of the same designation may or may not carry additional meaning, This is important, because SBVR examples assume that plurals are conventional and irrelevant, but the Date Time Vocabulary says that the use of verb tenses in natural language conveys indexical time intent. That is: 'John is in London' and 'John was in London' have different meanings in English. Do they have different meanings in SBVR SE? And if so, do they always have different meanings? Natural language convention requires that a statement that dates a past event uses the past tense, e.g., 'John was in London in 2008.' Is it meaningful in SBVR SE to say (in 2012) 'John is in London in 2008'? And does that mean a different proposition from 'John was in London in 2008'?
Title: Clarify the Purpose and Scope of SBVR, the Authority for SBVR Vocabulary Content, and SBVR Vocabularies Do Not Include Business Instance Data Source: Business Semantics Ltd, Donald Chapin, (Donald.Chapin@btinternet.com) Summary: Since SBVR v1.0 was published in January 2008 there has been widespread misinterpretation and misrepresentation of SBVR as a data modeling specification that is not easy to refute with finality because Clause 1 “Scope’ does not make it clear that the authority for the content of an SBVR Vocabulary is the usage of terms and other designations in a corpus of business documentation. Further contributing to the problem is the fact that the Subclause 10.1 formal semantics for SBVR is one that is based on a fact-oriented data modeling paradigm. Even though the formal interpretation is meant to be specified only in terms of formal logic there is wide reference to “facts”. Since the representations of facts are what data is, without statements to the contrary this can be used as a basis for incorrectly interpret the SBVR vocabularies in Clause 7, 8, 9, 1 & 12 as a collection of vocabularies for fact-oriented data modeling rather than documentation of the business language used by business people. Resolution: 1. Clarify the Scope of SBVR in Clause 1 to be explicit that SBVR does not include business instance data; and make it clear that the content of an SBVR vocabulary documents the meaning of terms that business authors intend when they use them in their business communications, as evidenced in their written documentation, especially governance documentation. 2. Add a list of purposes / uses of SBVR 3. Explain that “Semantic Anchors” are the best way to relate SBVR vocabularies to data models and models for reasoning over data. 4. Make it clear that SBVR vocabularies are different from all forms of data models models for and reasoning over data.. 5. Make fact an abstract concept in Clause 13.2.2 as instances of business facts (instance data) and fact statements do not go into an SBVR Vocabularies or Rulebooks. 6. Clean up miscellaneous uses of the word “fact”. Revised Text: … to follow
Title: Fix the anomaly in the subcategory structure of ‘concept’ to include ‘individual verb concept’ in SBVR Source: RuleML Initiative, John Hall, (john.hall@modelsystems.co.uk) Summary: SBVR handles noun concepts and verb concepts asymmetrically: • ‘concept’ generalizes ‘noun concept’ and ‘verb concept’ • ‘noun concept’ generalizes ‘general concept’ and ‘individual concept’ – i.e. ‘general concept’ means ‘general noun concept’ and ‘individual concept’ means ‘individual noun concept’ There are no equivalents for ‘verb concept’. SBVR does not explicitly define ‘individual verb concept’, so cannot say: • ‘individual concept’ generalizes ‘individual noun concept’ and ‘individual verb concept’ (inheriting from: ‘concept’ generalizes ‘noun concept’ and ‘verb concept’) • ‘verb concept’ generalizes ‘general verb concept’ and ‘individual verb concept’ (paralleling: ‘noun concept’ generalizes ‘general noun concept’ and ‘individual noun concept’) If it did, this structural inconsistency would be removed. It would also be helpful in using SBVR. Individual noun concepts, such as “EU-Rent” and “Luxembourg”, are useful in defining bodies of shared meanings in SBVR. If SBVR included ‘individual verb concept’, an SBVR body of shared meanings could include individual verb concepts such as “EU-Rent is incorporated in Luxembourg”. Resolution: 1. Change the preferred term that is currently ‘individual concept’ to ‘individual noun concept’ to clarify that it applies to noun concepts only 2. Add the concept ‘individual verb concept’ for a proposition that is a Clause 8 verb concept with all its roles quantified (closed) by individual (noun) concepts to fix the anomaly in the subcategory structure of ‘concept’. Revised Text: On printed page 22 in Clause 8.1.1 REPLACE the current term heading “individual concept” WITH “individual noun concept” And REPLACE “concept”, the first term in the definition, WITH “noun concept” On printed page 27 in Clause 8.1.2 at the end of the clause ADD this entry for ‘individual verb concept’: individual verb concept Definition: proposition that is based on exactly one verb concept in which each verb concept role is filled by an individual noun concept Note: … some explanatory comments Example: … some illustrative examples REPLACE the signifier “individual concept” WITH “individual noun concept” in the following places (but not in the “Source” subentry reference to ISO 1087-1 in entry for the concept current termed “individual concept’) • … to be identified and added REPLACE the following diagrams WITH diagrams that repolace the signifier “individual concept” with “individual noun concept”: • Figure 8.1 • Figure 9.3 • Figure 11.2 … plus fixes for any additional side effects:
Problem: If "body of shared concepts" were defined as [the set of] all of the concepts within a body of shared meanings", then I dont think the following entry would be needed: "body of shared concepts includes concept". Resolution: 1. Change the definition of "body of shared concepts" to: the set of all of the concepts within a body of shared meanings" 2. Eliminate the entry: body of shared concepts includes concept
Problem: The current definition of "representation uses vocabulary" is "the representation is expressed in terms of the vocabulary". I think the un-styled "term" (in terms of) is a bad choice for the definition. A better choice might be based on. Resolution: Change the definition of "representation uses vocabulary" to: "the representation is expressed based on the vocabulary".
All, The re-sequencing of Clause 11 has initially proven quite worthwhile, so I have been encouraged to do similar work on Clause 8. As before, I made no changes to the entries themselves whatsoever. (If I did, it was purely an error and should be corrected. Also, my work should be double-checked for any entries inadvertently omitted.) I used a Word version kindly supplied by Linda Heaton, which I believe is from the latest convenience document. (It does have some styling problems, which I have noted.) I hope we can move forward with this revision expeditiously. By the way, I found this re-sequencing much harder than Clause 11, which I am much more familiar with. Ron ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Problem Statement: SBVR adheres to ISO 1087 as much as possible. Clearly evident in the structure of ISO 1087 is that no definition should appear until all terms within it have been defined (as needed). This best practice (rule) results in a logical, easy-to-follow presentation. Clause 8 of SBVR is clearly broken in this regard. The result is significant lack of clarity, making detection of errors unnecessarily difficult. The possibility of misinterpretation (or non-comprehension) by software engineers and other readers is high. Resolution: Apply the ISO 1087 rule about sequencing vocabulary entries rigorously. This re-sequencing requires no changes in the entries themselves (but does suggest some). Two files containing all Clause 11 entries are attached. One file is unchanged except that entries are numbered. The other file is re-sequenced. The entry numbers reappear in the re-sequenced file indicating the original location of each entry. New subheadings are suggested.
There are two minor ambiguities in definitions of types of noun concept: 1. ‘unitary concept’ is defined as ‘individual concept or general concept that always has at most one instance’ . This is ambiguous because it is not clear whether ‘that always has at most one instance’ applies to both ‘individual concept’ and ‘general concept’ or only to ‘general concept’. 2. ‘individual concept’ is defined as ‘concept that corresponds to only one object [thing]’ (adopted from ISO 1087-1) This is ambiguous because it is not clear whether ‘only’ means ‘exactly one’ or ‘at most one’. The second note in the entry says “While each referring individual concept has at most one and the same instance …” suggesting that ‘only’ means ‘at most one’. Also, terms used for types of noun concept do not match their definitions. In SBVR, ‘concept’ includes both ‘noun concept’ and ‘verb concept’, but some terms use ‘concept’ for ‘noun concept’. For example, the definition for ‘general concept’ is for a specialization of ‘noun concept’. Discussion: The terms for types of noun concept became a concern after ‘fact type’ was replaced by ‘verb concept’ in Clause 8. Resolution: Update the definitions of ‘unitary concept’ and ‘individual concept’ to remove the ambiguities. Throughout the specification, replace the terms ‘general concept’, ‘unitary concept’ and ‘individual concept’ with, respectively, ‘general noun concept’, ‘unitary noun concept’ and ‘individual noun concept’ Revised Text: On printed page 21 in Clause 8.1.1 REPLACE unitary concept Definition: individual concept or general concept that always has at most one instance General Concept: noun concept WITH unitary noun concept Definition: general noun concept that always has at most one instance or individual noun concept On printed page 22 in Clause 8.1.1 REPLACE individual concept FL Source: ISO 1087-1 (English) (3.2.2) [‘individual concept’] Definition: concept that corresponds to only one object [thing] General Concept: unitary concept Concept Type: concept type Necessity: No individual concept is a general concept. Necessity: No individual concept is a verb concept role. WITH individual noun concept FL Source: based on ISO 1087-1 (English) (3.2.2) [‘individual concept’] Definition: noun concept that corresponds to at most one thing General Concept: unitary noun concept Concept Type: concept type Necessity: No individual noun concept is a general noun concept. Necessity: No individual noun concept is a verb concept role. UPDATE NOUN CONCEPT TERMS: REPLACE the signifier “general concept” WITH “general noun concept” … list of replacement locations to be provided REPLACE the signifier “unitary concept” WITH “unitary noun concept” everywhere REPLACE the signifier “individual concept” WITH “individual noun concept” everywhere except for the “Source” subentry reference to ISO 1087-1 in the entry for the concept currently termed “individual concept’ UPDATE DIAGRAMS: REPLACE the following diagrams WITH diagrams that replace the signifiers “general concept”, “unitary concept” and “individual concept” with, respectively, “general noun concept”, “unitary noun concept” and “individual noun concept”: • Figure 8.1 • Figure 9.3 • Figure 11.2 • Diagram in Clause 13.4 on printed page 198
In clause 8.3.4, the term 'verb concept wording' is defined as: "representation of a verb concept by an expression that has a syntactic structure involving a signifier for the verb concept and signifiers for its verb concept roles" In the same clause, the term 'noun form' is defined as: "verb concept wording that acts as a noun rather than forming a proposition" One would expect therefore, that a noun form of a verb concept would be a gerund, such as 'car transfer' for 'branch1 transfers car to branch2', where the 'noun form' denotes the same actualities as the verb concept. But only the last Example (which is hard to understand because of a particularly bad choice of verb) is said to be about gerunds. The other examples clearly are not. The first Example is: "'transferred car of car transfer' for the verb concept 'car transfer has transferred car'. This form yields a transferred car." The instances of 'car transfer has transferred car' are actualities of a car being involved in a car transfer. But the cited text says the instances of the 'noun form' 'transferred car of car transfer' are cars, not actualities. Similarly, the interpretation of the other two examples of 'noun forms' correspond to numbers, not actualities. So the instances of a noun form of a verb concept need not be instances of the verb concept! The noun form therefore cannot be a 'verb concept wording'. The noun form does not represent the verb concept! It appears that there are two different concepts here. Noun form 1 is "verb concept wording that acts as a noun." That is the gerund in the last Example. In the other examples, the noun form represents a derived concept that is what SBVR calls a 'situational role'. The intent of 'noun form 2' is "representation of a situational role by an expression that has a syntactic structure involving a signifier for the verb concept that the role is derived from and signifiers for some of its verb concept roles". Finally, use of noun form 2 in declaring a glossary item for a situational role would be preferable to using only the role designation. In particular, the explicit appearance of other role placeholders in the noun form would permit them to be used directly in defining the situational role. For example: cardinality Definition: nonnegative integer that is the number of distinct elements in a given set or collection could be declared with the noun form: _cardinality_ of _set_ Definition: nonnegative integer that is the number of distinct elements in the set
Inadequate, Overlapping and Disorganized Specifications for Sets and Collections of Concepts, Meanings, and Representations Problem: Assumptions Two assumptions are basic to the eight points of this problem statement: • SBVR must provide a business vocabulary for business people and business analysts to talk clearly and precisely about terminological dictionaries and rulebooks and what they represent. • The various aspects of this Issue must be addressed holistically. They can be resolved only by unifying, normalizing and completing all related specifications. (Thus, the need for a new unifying Issue.) Problems 1. A known problem in SBVR is that the current version does not make clear what the fundamental unit of interoperability in SBVR is. No matter how that issue is resolved the unit should: • Be identifiable from a business point of view. • Not always have to be the full, non-redundant set of concepts, meanings, or representations. The existing content of Clause 11 does not currently provide an adequate term for the second of these. This Issue proposes “collection” for that purpose. Note: The term “collection” in the following discussion is never actually used on its own. Rather, it always appears with qualification – e.g., ‘collection of representations’. 2. Another known problem in SBVR centers on the use of the word “container” in e-mails and discussion. (Use of the signifier “container” per se is not part of this Issue.) It is unclear (to some) whether “container” refers to the ‘thing that contains’, to ‘what is contained’, or to both. The term is easily misused and misinterpreted. Also there are many variations of what is (or could be) contained (e.g., sets, collections, etc.). SBVR needs a precise, non-overlapping vocabulary for these things from a business point of view. 3. Another known problem in SBVR is that the existing content of Clause 11.2.2.3 “communication content” (a.k.a. “document content”) is not adequate for all purposes to which it might be put. SBVR needs a richer (but still minimal) set of concepts to address this area. 4. Certain existing terms in the existing content of Clause 11 (e.g., ‘terminological dictionary’ and ‘rulebook’) conflate ‘completeness and non-redundancy’ (i.e., being a set) with ‘primary purpose’ or ‘essence’. This conflation needs to be eliminated. In the real world for example, a rulebook does not have to be complete (e.g., it may contain only what is appropriate for a given audience), and it does not have to be non-redundant. It can contain the same rule statements in different sections, the intent being to provide the greatest clarity when being used by members of some speech community. 5. SBVR currently provides no means to talk about a collection of representations that is complete with respect to one or more specific concepts, but not complete with respect to all concepts in the body of shared meanings. Example: A listing of all baseball rules that address the concepts “strike” and “ball” only. 6. With respect to interoperability there is a minimum set of pragmatic business specifications (such as completeness, effective date, shelf life, mutability, etc.) needed for things communicated. SBVR does not currently support such specifications. Note: There is no intent or need to get into document management or rule management. The dividing line is this: SBVR does not get into organizational issues (e.g., author, sponsor, reviewer, etc.), workflow issues (e.g., status, pre-approval distribution, sign-off, impact assessment, etc.), motivation (rationale, goals, risks), etc. SBVR must, however, provide minimum viability criteria for any sets or collections communicated. Otherwise the communicated content is not really useful and trustworthy on the receiving end. Consequently the purpose of interoperability is defeated. 7. Certain kinds of collections relevant to inter-operability are missing from the current content of Clause 11 – most notably ‘record’ (not IT ‘records’). Proper incorporation of this and other kinds of collections is needed. 8. Issue 16103, which addresses “speech community representation”, needs to be worked into a holistic solution.
New SBV issue: Distinguishing the senses of infinitive and present tense From Don Baisley There are many verbs for which the present tense of a verb conveys a particularly different sense than the infinitive. The difference I refer to is not about "the present time", but about "occurring at least occasionally". For example, the statement that "Pam surfs" (present tense) combines the meaning of "to surf" (the infinitive) and the meaning that “it happens at least occasionally”. For such verbs, there is a challenge when using SBVR's typical pattern of defining verb concepts in the present tense. It tends to conflate the infinitive sense of a verb with the different sense meant by the present tense. That conflation causes problems. This is not an issue for ORM or other approaches that do not try to support natural language tense in a generic way. The problem has no apparent impact for many verbs where the present tense sense of "occurring at least occasionally" is inconsequential or inapplicable. The problem is especially troublesome for eventive verbs. Most SBVR verbs are stative, so the problem has tended to go unnoticed in the SBVR vocabulary itself. If supporting tense in a generic way, in logical formulations, the other tenses should be built on objectifications that start with the infinitive sense of a verb, not with the present tense. Also, modal operations like obligation build on the infinitive sense. For examples below, I define verb concept forms for generic "tense" concepts using the verb "occurs" (where the there is a role that ranges over the concept 'state of affairs'). The choices of signifier and form are arbitrary (not necessary), but seem to convey the sense of the tenses naturally. Example: 'person surfs' (present tense) 'person surf' (the infinitive sense) Where someone puts 'person surfs' in a business glossary, there is an underlying verb concept that has the sense of "to surf", the infinitive. I show it here in examples as 'person surf' (leaving out the infintizing "to"). This underlying verb concept is necessary to correctly formulate other tenses, and even necessary to formulate use of the present tense in some cases, which I will show later. Here are several examples of statements and interpretations using generic tense concepts built on the verb "occur". To be terse, I show objectification using brackets. Pam surfs. [Pam surf] occurs Pam is surfing. [Pam surf] is occurring Pam was surfing. [Pam surf] was occurring Pam has been surfing. [Pam surf] has been occurring Pam surfed. [Pam surf] occurred Pam will be surfing. [Pam surf] will be occurring Pam will surf. [Pam surf] will occur Pam will have been surfing. [Pam surf] will have been occurring The second example above, "Pam is surfing", can serve to illustrate the need to build on the infinitive rather than the present tense sense. To build on the present tense would be to say the thing that “is occurring” is Pam surfing at least occasionally, which is incorrect. The present continuous and other tenses do not include the present tense sense of occurring at least occasionally, so they cannot rightly be built upon a concept that conveys that sense. I said above I would show where the infinitive sense is sometimes needed even for the present tense. Here is a case where the infinitive 'person surf' concept is needed to formulate a statement that uses "surf" only in the present tense: Pam talks while she surfs. Wrong Interpretation I1: [Pam surfs] occurs while [Pam talks] occurs I1 misses the key sense of the statement, because [Pam surfs] (present tense) means that surfing is something Pam does at least occasionally and [Pam talks] means that talking is something that Pam does at least occasionally. I1 applies 'state of affairs1 occurs while state of affairs2 occurs' to the wrong states of affairs (the states in which Pam occasionally surfs and Pam occasionally talks). Right Interpretation I2: [[Pam surf] occur while [Pam talk] occur] occurs I2 correctly factors out the tense and applies it at an outer level (as we often do with modal operations). The conjunction joins objectifications of the underlying sense of "to surf" and "to talk" without the added meaning of the present tense (that the surfing or talking is at least occasional). The sense of present tense (happening at least occasionally) is then added at the outside where it applies to the simultaneous actions. SBVR does not prevent verbs concepts from being defined in glossaries in the infinitive , as is typical of dictionary definitions of verbs. That approach has always been available. But that approach is not used in SBVR’s own glossary and examples. In general, the sense of “occurs at least occasionally” is absent from SBVR’s own verb concepts, so the distinction is unimportant. But business rules and facts run into the problem. E.g., a EU-Rent rule about whether a renter smokes vs. a rule about whether he is smoking when in a rental car. Recommendation: It will be best to resolve this in a way that does not disturb the business-friendly approach of showing verb concept readings in the present tense. It might be possible to define a pattern in SBVR Structured English by which verb concepts with an infinitive sense are implied where present tense versions are explicitly presented in a glossary. Examples of formulations need to show the distinction. Existing examples should be examined and fixed as needed. New formulation examples might be helpful to demonstrate using generic tense concepts to build on a root verb concept. None of this changes the meaning of 'state of affairs' or 'objectification', but understanding this issue and its solution might help bring clarity to some of the examples that have been discussed.
The definitions in most of the model entries in Clause 8.1.2 do not align with Clause 9 and Clause 10 modal definitions. Resolution: Align the definitions in Clause 8, 9 & 10 by changing the definitions in Clause 8; adding an intermediate concept to make the definition of “proposition is permitted to be true” intelligible to business people; and adding a definition for “actual world” to Clause 10. Revised Text: REPLACE the Definition in the entry for “proposition is necessarily true” in Clause 8.1.2 on printer page 26: Definition: the proposition always corresponds to an actuality WITH: Definition: the proposition corresponds to an actuality in all possible worlds REPLACE the Definition in the entry for “proposition is possibly true” in Clause 8.1.2 on printer page 26: Definition: it is possible that the proposition corresponds to an actuality WITH: Definition: the proposition corresponds to an actuality in some possible world Add a new Entry after the entry for “proposition is obligated to be true” in Clause 8.1.2 on printer page 26: proposition is obligated to be false Definition: the proposition does not correspond to an actuality in any acceptable world REPLACE the Definition in the entry for “proposition is permitted to be true” in Clause 8.1.2 on printer page 27: Definition: the proposition corresponds to an actuality in at least one acceptable world. WITH: Definition: the proposition is not obligated to be false REPLACE the Definition in the entry for “permissibility formulation” in Clause 9.2.4 on printer page 57: Definition: modal formulation that formulates that the meaning of its embedded logical formulation is true in some acceptable world WITH: Definition: modal formulation that formulates that the meaning of its embedded logical formulation is permitted to be true ADD immediately after the entry for “acceptable world” in Clause 10.1.2 on printed page 111 the following new entry: actual world Definition: the possible world that is taken to be actual for some purpose, in particular, for the conduct of business and the application of business rules Note: the actual world is a set of things, situations and facts about them that some person or organization takes to be true for some purpose. In most cases, it is the best estimate of the actual state of the world that is of interest at a particular time.
ftp://ftp.omg.org/pub/issue-attachments/17599/17599.doc
SBVR v1.1 Clause 8 says: Note: in the glossary entries below, the words “Concept Type: role” indicate that a general concept being defined is a role. Because it is a general concept, it is necessarily a situational role and is not a verb concept role. How does one declare an attributive role that is not a general concept? SBVR v1.1 appears to use such declarations to also declare roles that are attributive roles of a given noun concept and thus also in the attributive namespace of the noun concept. For example, clause 8.6 declares 'cardinality', which is an attributive role of integers with respect to 'sets', in a glossary entry with Concept type: role. But 'cardinality' is not a general concept; nothing is a 'cardinality', full stop. An integer can only be a 'cardinality' OF something. it is a purely attributive term. As a term for a general concept, 'cardinality' is thus a term in the Meaning and Representation namespace; it has no 'context'. The problem arises in defining attributive roles of general noun concepts, such as 'occurrence has time span' and 'schedule has time span', where the definitions of the two roles are importantly different because they are attributes of different general concepts that are only similar in nature. Neither is a situational role. That is, neither is a general concept. No time interval is a 'time span', full stop. A time interval must be a time span OF something. One 'time span' is in the attributive namespace of 'schedule', and a different 'time span' designation is in the attributive namespace of 'occurrence'. Neither is in the DTV.Situations vocabulary namespace per se. How can this be declared using SBVR conventions? Declaring them both in glossary entries with Concept Type: role apparently makes them conflicting designations for 'situational roles' in the DTV.Situations vocabulary. Does simply declaring the verb concept 'occurrence has time span' declare the attributive role? If so, how is the range of the role declared? And where does the definition of the attributive role go?
There are four small but inportant wording additions need to clarify three entries in the Clause 10.1.2: • In “possible world” and “universe of discourse” the word “object” has the signifier “thing” in sBVR • Make it clear that the “at some point in time” is the “present time of the possible word” as set forth in SBVr Clause 10.1.1. • The referents of “corresponding propositions or states of affairs” at the of the definition for ‘state of affairs’ is not clear. Resolution: Make the clarifications as identifed in Issue Summary. Revised Text: ADD in the second sentence in the Note in the entry for “possible world” in Clause 10.1.2 on printed page 117 after the phrase “any given set of objects”: [things] ADD to the end of the last sentence in the Note in the entry for “possible world” in Clause 10.1.2 on printed page 117: Thus, in the context of a static constraint declared for a given business domain, a “possible world” would correspond to (but not be identical to) a state of the domain’s fact model that could exist at some point in time. the following text: , which is the “present time” of the possible world.“ ADD the word “respectively” at the end of the Definition in the entry for “state of affairs” in Clause 10.1.2 on printed page 119 after the phrase “set of objects”: ADD at the beginning of the Definition in the entry for “universe of discourse” in Clause 10.1.2 on printed page 120 after the phrase “set of objects”: [things]
SBVR is intended for development of semantic models of businesses (and enterprises run on similar lines, such as public sector bodies and not-for-profit organizations). Its scope says “This specification defines the vocabulary and rules for documenting the semantics of business vocabularies, business facts, and business rules”. A lot of SBVR RTF email and teleconference discussion seems to be taken up with examples that are at best tenuously related to business, and often not at all related to business. There is, of course, no reason that people should not use SBVR as SVR – Semantics of Vocabulary and Rules – for any universe of discourse, whether business-related or not. But it is important to keep focus on what SBVR is intended for. Dependencies with other Issue Resolutions: None Discussion: There are two aspects of keeping SBVR’s focus on business: 1. The context of an SBVR model of a business – a body of shared concepts, represented as one or more terminological dictionaries and rulebooks – is the actual world in which the business operates. 2. The content of an SBVR model of a business is the meanings of the definitions of relevant things, relationships and guidance in the business. The universe of discourse is the part of the business selected by the business owners to be within scope. For example, in EU-Rent (as used in the SBVR specification) it is car rentals as opposed to finance, car purchasing and sales, premises management, HR, etc. This issue is about a matter of SBVR practice and can be addressed with notes (or perhaps in more general editorial). Resolution: Add notes under the entry for ‘body of shared meanings’: • To describe the universe of discourse modeled by the body of shared meanings • To emphasize that the body of shared meanings comprises only meanings Revised Text: In 11.1.1.2, under the entry for body of shared meanings, add the following notes: Note: When modelling a business (such as EU-Rent), the universe of discourse is bounded by what the business owners decide is in scope. That would be the actual world of some part of EU-Rent’s business (e.g. rentals, as opposed to, say, premises management, purchase/sales of cars, or HR) and some possible worlds that are reachable from the actual world. If the EU-Rent owners say that they are considering renting RVs or starting up in China, then possible worlds that include these kinds of business are in the universe of discourse. If EU-Rent is not considering renting construction equipment or camping gear, then possible worlds that include these kinds of business are not in the universe of discourse – and neither are possible worlds that include impossibilities. Whether ‘Kinnell Construction rented backhoe 123 on 2012-08-28’ or ‘John rode into work on a unicorn’ correspond to states of affairs or not, are not relevant to EU-Rent. They are out of scope. In-scope propositions may have to be constrained by necessities to ensure that they are not impossible. e.g. ‘Necessity: Each rental car is stored at at most one branch [at any given time].’ Note: A body of shared meanings contains meanings of: • noun concepts that define kinds of thing in the universe of discourse • verb concepts that define relationships between kinds of thing in the universe of discourse • elements of guidance that constrain or govern the things and relationships defined by the concepts. It does not contain ground facts or facts derived from ground facts (other than as illustrative examples), or things in the universe of discourse, or information system artifacts that model things in the universe of discourse – although it may provide vocabulary to refer to them.
Title: Fix the anomaly in the subcategory structure of ‘concept’ to include ‘individual verb concept’ in SBVR Source: RuleML Initiative, John Hall, (john.hall@modelsystems.co.uk) Summary: SBVR handles noun concepts and verb concepts asymmetrically: • ‘concept’ generalizes ‘noun concept’ and ‘verb concept’ • ‘noun concept’ generalizes ‘general concept’ and ‘individual concept’ – i.e. ‘general concept’ means ‘general noun concept’ and ‘individual concept’ means ‘individual noun concept’ There are no equivalents for ‘verb concept’. SBVR does not explicitly define ‘individual verb concept’, so cannot say: • ‘individual concept’ generalizes ‘individual noun concept’ and ‘individual verb concept’ (inheriting from: ‘concept’ generalizes ‘noun concept’ and ‘verb concept’) • ‘verb concept’ generalizes ‘general verb concept’ and ‘individual verb concept’ (paralleling: ‘noun concept’ generalizes ‘general noun concept’ and ‘individual noun concept’) If it did, this structural inconsistency would be removed. It would also be helpful in using SBVR. Individual noun concepts, such as “EU-Rent” and “Luxembourg”, are useful in defining bodies of shared meanings in SBVR. If SBVR included ‘individual verb concept’, an SBVR body of shared meanings could include individual verb concepts such as “EU-Rent is incorporated in Luxembourg”. Resolution: 1. Change the preferred term that is currently ‘individual concept’ to ‘individual noun concept’ to clarify that it applies to noun concepts only 2. Add the concept ‘individual verb concept’ for a proposition that is a Clause 8 verb concept with all its roles quantified (closed) by individual (noun) concepts to fix the anomaly in the subcategory structure of ‘concept’. Revised Text: On printed page 22 in Clause 8.1.1 REPLACE the current term heading “individual concept” WITH “individual noun concept” And REPLACE “concept”, the first term in the definition, WITH “noun concept” On printed page 27 in Clause 8.1.2 at the end of the clause ADD this entry for ‘individual verb concept’: individual verb concept Definition: Definition to be replaced proposition that is based on exactly one verb concept in which each verb concept role is filled by an individual noun concept … some explanatory comments Example: … some illustrative examples REPLACE the signifier “individual concept” WITH “individual noun concept” in the following places (but not in the “Source” subentry reference to ISO 1087-1 in entry for the concept current termed “individual concept’) • … to be identified and added REPLACE the following diagrams WITH diagrams that repolace the signifier “individual concept” with “individual noun concept”: • Figure 8.1 • Figure 9.3 • Figure 11.2 … plus fixes for any additional side effects
The DTV RTF has pointed out the value of adding to SBVR a very generic concept for all kinds of occurrences so that all specifications that define a particular kind of occurrence, e.g. occurrence in time, occurrence in space, can be consistent if they adopt and specialize the SBVR generic occurrence concept. This approach also provides the ability of specifications that deal with occurrence to constrain the generic concepts adopted from SBVR to fit their specification. Resolution: 1. Incorporate that a state of affairs is not a meaning in its definition. 2. Add a generic, overarching ‘occurrence’ noun concept. 3. Add a “what happens” noun concept that is a role of ‘state of affairs’. 4. Add a verb concept that defines the multiple relationship between “what happens’ and ‘occurrence’. 5. Fix the definiiton of ‘state of affairs is actual’ 6. Clarify the Note for ‘actuality’. 7. Remove confusing and unnecessary wording in the entry for ‘situation’. Revised Text: REPLACE Figure 8.8 in Clause 8.5 on printed page 40 WITH: In clause 8.5, in the entry for 'state of affairs', REPLACE the Definition: Definition: event, activity, situation, or circumstance with Definition: res that is an event, activity, situation, or circumstance In clause 8.5, immediately before the entry for 'state of affairs is actual', INSERT three new entries: occurrence Definition: state of affairs that is the happening of another state of affairs for a given time interval and/or at a given location and/or in some other dimension what-happens state of affairs Definition: state of affairs that can happen for a given time interval and/or at a given location and/or in some other dimension what-happens state of affairs has occurrence Definition: the occurrence is the realization of the state of affairs In clause 8.5, in the entry for 'state of affairs is actual', REPLACE the existing Definition: Definition: the state of affairs happens (i.e., takes place, obtains) with: Definition: the state of affairs is happening (i.e., takes place, obtains) In clause 8.5, in the entry for 'actuality', REPLACE the Note: Note: Actualities are states of affairs that actually happen, as distinct from states of affairs that don’t happen but nevertheless exist as subjects of discourse and can be imagined or planned. with: Note: Actualities are states of affairs that are actually happening, as distinct from states of affairs that are not happening but nevertheless exist as subjects of discourse and can be imagined or planned. In clause 11.1.5.2, in the entry for 'situation' on printed page 154, REMOVE • The phrase “that provides the context from which roles played may be understood or assessed” at the end of the Definition as it is about purpose and not essential meaning. • the words “a state of affairs” at the end of the first Dictionary Basis. ADD two noun concepts, ‘occurrence’ and ‘what-happens state of affairs’, to the following line in the paragraph beginning with “The classes in the metamodel that mirror …” in 13.2.2 “MOF Classes for SBVR Noun Concepts”: Clause 8: meaning, concept, expression, state of affairs, actuality, thing, set ADD two noun concepts, ‘property’ and ‘viewpoint’, to the following line in the paragraph beginning with “The classes in the metamodel that mirror …” in 13.2.2 “MOF Classes for SBVR Noun Concepts”: Clause 11: community, situation, res