Issue 9933: SBVR Issue - Context for understanding representations (sbvr-ftf) Source: Escape Velocity (Mr. Don Baisley, donbaisley(at)live.com) Nature: Uncategorized Issue Severity: Summary: Context for understanding representations A definition given for a fact type commonly refers to the roles of the fact type using placeholders from a form of expression of the fact type. But the SBVR vocabulary provides no relationship between a definition and the form of expression for which the definition makes sense (the one having the placeholders used in the definition). A new fact type must be added to SBVR in order to capture this relationship. Otherwise there is insufficient context to interpret the definition. The problem occurs because there can be many forms of expression for a fact type, and these can uses different designations for placeholders. A similar problem occurs where a note or example for a concept only makes sense in relation to a particular designation or form of expression, such as when the example or note refers to the designation or a placeholder of the form of expression. Similarly, the expression of a reference supporting a concept often does not include the designation used in a source document because the reference is given in relation to that particular designation for the concept. But if there are many designations for the concept, the reference cannot be adequately interpreted with respect to the source document. Also, a note about a rule might make sense only in the context of a particular statement of the rule because the note refers to specific words used in the rule. But the same rule can be represented by many different statements. Recommendation: Add a new fact type to the Meaning and Representation Vocabulary after ‘representation represents meaning’: representation relates to representation And explain that the first representation is understood in the context of the second representation. Resolution: Deferred to first SBVR Revision Task Force because we ran out of time. Revised Text: Actions taken: July 20, 2006: received issue Discussion: This Issue was received just before the Issue deadline which was only 6 weeks before the FTF report was due, and more basic Issues had to be resolved first. End of Annotations:===== ubject: SBVR Issue - Context for understanding representations Date: Thu, 20 Jul 2006 12:57:24 -0700 X-MS-Has-Attach: yes X-MS-TNEF-Correlator: Thread-Topic: SBVR Issue - Context for understanding representations Thread-Index: AcasNrfkOrvGFiubRXG/XCz4jIfvuQ== From: "Baisley, Donald E" To: , "Juergen Boldt" X-OriginalArrivalTime: 20 Jul 2006 19:57:24.0376 (UTC) FILETIME=[B81D0980:01C6AC36] Context for understanding representations A definition given for a fact type commonly refers to the roles of the fact type using placeholders from a form of expression of the fact type. But the SBVR vocabulary provides no relationship between a definition and the form of expression for which the definition makes sense (the one having the placeholders used in the definition). A new fact type must be added to SBVR in order to capture this relationship. Otherwise there is insufficient context to interpret the definition. The problem occurs because there can be many forms of expression for a fact type, and these can uses different designations for placeholders. A similar problem occurs where a note or example for a concept only makes sense in relation to a particular designation or form of expression, such as when the example or note refers to the designation or a placeholder of the form of expression. Similarly, the expression of a reference supporting a concept often does not include the designation used in a source document because the reference is given in relation to that particular designation for the concept. But if there are many designations for the concept, the reference cannot be adequately interpreted with respect to the source document. Also, a note about a rule might make sense only in the context of a particular statement of the rule because the note refers to specific words used in the rule. But the same rule can be represented by many different statements. Recommendation: Add a new fact type to the Meaning and Representation Vocabulary after .representation represents meaning.: representation relates to representation And explain that the first representation is understood in the context of the second representation. Don Baisley | Consulting Engineer | Systems & Technology Unisys | 25725 Jeronimo Road | Mission Viejo, CA 92691 | 949-380-6382 THIS COMMUNICATION MAY CONTAIN CONFIDENTIAL AND/OR OTHERWISE PROPRIETARY MATERIAL and is thus for use only by the intended recipient. If you received this in error, please contact the sender and delete the e-mail and its attachments from all computers. Subject: RE: issue 9933 -- SBVR FTF issue Date: Wed, 11 Oct 2006 14:45:45 -0700 X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: issue 9933 -- SBVR FTF issue Thread-Index: AcasPGXGZdKw9433Rlupir+7B6NCvRBQPUxg From: "Baisley, Donald E" To: X-OriginalArrivalTime: 11 Oct 2006 21:45:45.0726 (UTC) FILETIME=[9B8191E0:01C6ED7E] After looking further at ISO.s use of .context. and its primary definition in my dictionary, I think we can solve this issue by adding the fact type .representation has context. along with the role .context.. context Definition: representation that occurs before or around another particular representation and that helps to discern its full meaning. Concept Type: role representation has context Regards, Don -------------------------------------------------------------------------------- From: Juergen Boldt [mailto:juergen@omg.org] Sent: Thursday, July 20, 2006 1:34 PM To: issues@omg.org; sbvr-ftf@omg.org Subject: issue 9933 -- SBVR FTF issue This is issue # 9933 From: "Baisley, Donald E" SBVR Issue - Context for understanding representations Context for understanding representations A definition given for a fact type commonly refers to the roles of the fact type using placeholders from a form of expression of the fact type. But the SBVR vocabulary provides no relationship between a definition and the form of expression for which the definition makes sense (the one having the placeholders used in the definition). A new fact type must be added to SBVR in order to capture this relationship. Otherwise there is insufficient context to interpret the definition. The problem occurs because there can be many forms of expression for a fact type, and these can uses different designations for placeholders. A similar problem occurs where a note or example for a concept only makes sense in relation to a particular designation or form of expression, such as when the example or note refers to the designation or a placeholder of the form of expression. Similarly, the expression of a reference supporting a concept often does not include the designation used in a source document because the reference is given in relation to that particular designation for the concept. But if there are many designations for the concept, the reference cannot be adequately interpreted with respect to the source document. Also, a note about a rule might make sense only in the context of a particular statement of the rule because the note refers to specific words used in the rule. But the same rule can be represented by many different statements. Recommendation: Add a new fact type to the Meaning and Representation Vocabulary after .representation represents meaning.: representation relates to representation And explain that the first representation is understood in the context of the second representation. Juergen Boldt Director, Member Services Object Management Group 140 Kendrick St Building A Suite 300 Needham, MA 02494 USA tel: +1 781 444 0404 x 132 fax: +1 781 444 0320 email: juergen@omg.org www.omg.org Date: Thu, 12 Oct 2006 19:03:32 -0400 From: Ed Barkmeyer Reply-To: edbark@nist.gov Organization: NIST User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.2) Gecko/20040804 Netscape/7.2 (ax) X-Accept-Language: en, fr, de, pdf, it, nl, sv, es, ru To: "Baisley, Donald E" , SBVR-FTF Subject: Re: issue 9933 -- SBVR FTF issue X-MailScanner-Information: Please contact postmaster@mel.nist.gov for more information X-MailScanner: Found to be clean X-MailScanner-SpamCheck: X-MailScanner-From: edbark@nist.gov X-Spam-Status: No Baisley, Donald E wrote: After looking further at ISO's use of "context" and its primary definition in my dictionary, I think we can solve this issue by adding the fact type 'representation has context' along with the role 'context'. context Definition: representation that occurs before or around another particular representation and that helps to discern its full meaning. Concept Type: role representation has context I have several different kinds of problems with this. First, 'context', with this definition, is only a Role in the context of this new fact type. I don't think SBVR should *ever* define a Fact Type Role separately from the Fact Type. When we structure the text the way Don proposes here, we are effectively agreeing to use the term 'context' ONLY to refer to this role in this fact type. And since that is not true, we should not do this this way. Second, we need to distinguish between an Expression and an occurrence of an Expression. The Expression "representation" occurs as a glossary item in Clause 8 and also occurs as a role/placeholder in Don's proposed fact type. If context is important to determining the Meaning of an Expression, then a Representation associates an *occurrence of* an Expression with a Meaning. (Alternatively, we can define an Expression to be a specific speech act, e.g., an occurrence of a Text.) Third, I don't understand how a Representation has a context. The occurrence of the Expression has a context, and one uses that context in determining the Meaning to associate with that occurrence. But once you have established the Representation relationship, the context is implicit in the associated Meaning. That is, the context of a Representation is a relationship between its Meaning and other Meanings. It happens that those other Meanings are Represented by the surrounding Expression occurrences, but that information is no longer useful. Using the example above, the interesting Representation associates the occurrence of the Expression "representation" with the Meaning "placeholder for first Role in the Fact type 'representation has context'". By comparison, the Representation of the occurrence of "representation" in Clause 8 associates it with the Meaning "individual Glossary Item in the SBVR Basic Vocabulary". In both cases, the semantic aspect of "context" is captured in that Meaning. Fourth, the physical context of an Expression occurrence is a structure of Expression occurrences. Is it important to capture that? And the semantic context of an Expression occurrence is the Meanings of the elements of that structure. So we can say that the semantic context of an Expression occurrence is a structure of Representations. Is this issue really about capturing the semantic context of Expression occurrences, as opposed to Representations? Finally, I'm still trying to figure out how this Issue relates to semantic formulations. An SBVR Definition is an Expression that should have a unique occurrence (as that Definition). And it is a Representation in the sense that it has an associated Meaning. The question is: What are the rules for formulating the associated Meaning? There seem to be 3 cases: 1) Essentially none of the sub-Expressions (signs, words) in the Definition Expression is a Vocabulary entry. In this case we cannot formulate that Meaning as anything other than the Definition Expression itself - the Text. 2) Essentially all of the sub-Expressions are SBVR keywords or vocabulary entries. In this case, we should be able to create a complete Semantic Formulation of the Meaning of the Definition Expression per Clause 9. In that formulation, the referents of the sub-Expressions will be unambiguous, because the parser understands the context of reference of the sub-Expressions. References to Roles that are unique to Fact Types will be properly captured and distinguished from references to Concepts that may have the same orthography, and so forth. Representation captures these associations. 3) Many of the sub-Expressions are SBVR keywords or vocabulary entries, and the rest is simple natural language text. One possibility is that the parser just converts this case to case (1). One possibility is that the parser can understand the parts of speech in the natural language text, per the reference Dictionary, and converts this case to case (2). And the final possibility is that the parser calls out the recognized sub-Expressions in and among the rest of the Definition Text and associates them with vocabulary entries and other semantic formulations. In this last case, the "semantic formulation" needs to capture the physical structure of the Expressions, and thereby the physical context, and it needs to capture the Representation (Meaning) of the recognized Expressions. The bottom line is that there are - Texts that you don't parse, - Texts that you parse and interpret, and - Texts that you parse only into mixed structures of unparsed Texts and interpreted texts (Representations). We need semantic formulations for the second case and physical context relationships for the last case, but we only use "semantic context" in the process of doing interpretation. This is issue # 9933 From: "Baisley, Donald E" SBVR Issue - Context for understanding representations A definition given for a fact type commonly refers to the roles of the fact type using placeholders from a form of expression of the fact type. But the SBVR vocabulary provides no relationship between a definition and the form of expression for which the definition makes sense (the one having the placeholders used in the definition). A new fact type must be added to SBVR in order to capture this relationship. Otherwise there is insufficient context to interpret the definition. The problem occurs because there can be many forms of expression for a fact type, and these can uses different designations for placeholders. A similar problem occurs where a note or example for a concept only makes sense in relation to a particular designation or form of expression, such as when the example or note refers to the designation or a placeholder of the form of expression. Similarly, the expression of a reference supporting a concept often does not include the designation used in a source document because the reference is given in relation to that particular designation for the concept. But if there are many designations for the concept, the reference cannot be adequately interpreted with respect to the source document. Also, a note about a rule might make sense only in the context of a particular statement of the rule because the note refers to specific words used in the rule. But the same rule can be represented by many different statements. Recommendation: Add a new fact type to the Meaning and Representation Vocabulary after 'representation represents meaning': representation relates to representation And explain that the first representation is understood in the context of the second representation. -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 FAX: +1 301-975-4482 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." Subject: RE: issue 9933 -- SBVR FTF issue Date: Fri, 13 Oct 2006 14:46:46 -0700 X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: issue 9933 -- SBVR FTF issue Thread-Index: AcbuUqdfHvp/aV3gQPSyNyPJc898lwAuRpCw From: "Baisley, Donald E" To: "SBVR-FTF" X-OriginalArrivalTime: 13 Oct 2006 21:46:46.0736 (UTC) FILETIME=[14B27500:01C6EF11] X-MIME-Autoconverted: from quoted-printable to 8bit by amethyst.omg.org id k9DLhV9Z005370 Ed wrote: > First, 'context', with this definition, is only a Role in > the context of this new fact type. I don't think SBVR should ... I agree with Ed. I think we should define 'context' in more general terms and then give a specific definition to the fact type 'representation has context'. Here is what I find for 'context' at dictionary.com: 1. the parts of a written or spoken statement that precede or follow a specific word or passage, usually influencing its meaning or effect: You have misinterpreted my remark because you took it out of context. 2. the set of circumstances or facts that surround a particular event, situation, etc. Other dictionaries seem to have the same two meanings in the same order. It seems to me that the first meaning is a specialization of the second, so I am happy to have "context" defined along the lines of either one of these. Can we have the more general concept be 'state of affairs' which seems to be a more general concept of 'circumstance' and which is also a more general concept of 'statement', 'definition' etc. if we resolve 9932 as requested (through 'actuality' being a more general concept of 'representation'). Then we can define the needed fact type: representation has context Definition: the context occurs before or around the representation and helps to discern its full meaning The text "the car has a full tank of gas" has different meanings in different contexts. Example 1 (defining a fact type): [car] is fueled Definition: the car has a full tank of gas Example 2 A car with license number 98765 is ready for renting. The car has a full tank of gas. Here we see the same expression used in two different representations. In the first case the context is the fact type form "[car] is fueled". In the second case the context is the statement "A car with license number 98765 is ready for renting." In each case, the context helps to understand the full meaning of the representation. The meaning of the first example is a fact type and the meaning of the second is a proposition. Currently, SBVR gives us words to relate each of the two representations to the one text and to their respective meanings. But SBVR does not yet give us words to relate the representations to their respective contexts. Adding 'context' and 'representation has context' is what is needed. Note that not all SBVR enabled tools will parse rules and definitions to create semantic formulations. If tools that don't are going to pass information via XMI to a tool that does, then they will need to be able to communicate context. Otherwise the tool that parses won't have what it needs to make sense out of the texts of statements and definitions. That is the practical reason for making this addition. Regards, Don Date: Mon, 16 Oct 2006 15:12:52 -0400 From: Ed Barkmeyer Reply-To: edbark@nist.gov Organization: NIST User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.2) Gecko/20040804 Netscape/7.2 (ax) X-Accept-Language: en, fr, de, pdf, it, nl, sv, es, ru To: "Baisley, Donald E" CC: SBVR-FTF Subject: Re: issue 9933 -- SBVR FTF issue X-MailScanner-Information: Please contact postmaster@mel.nist.gov for more information X-MailScanner: Found to be clean X-MailScanner-SpamCheck: X-MailScanner-From: edbark@nist.gov X-Spam-Status: No Baisley, Donald E wrote: I agree with Ed. I think we should define 'context' in more general terms and then give a specific definition to the fact type 'representation has context'. Do we need to define 'context'? I was hoping that we would just use the term in stating the fact-type: 'expression' has context 'representation' or something like that. Here is what I find for 'context' at dictionary.com: 1. the parts of a written or spoken statement that precede or follow a specific word or passage, usually influencing its meaning or effect: You have misinterpreted my remark because you took it out of context. 2. the set of circumstances or facts that surround a particular event, situation, etc. Other dictionaries seem to have the same two meanings in the same order. It seems to me that the first meaning is a specialization of the second, so I am happy to have "context" defined along the lines of either one of these. Can we have the more general concept be 'state of affairs' which seems to be a more general concept of 'circumstance' and which is also a more general concept of 'statement', 'definition' etc. if we resolve 9932 as requested (through 'actuality' being a more general concept of 'representation'). I agree with all of this. And in any case this is the meaning of the 'has context' fact-type. Then we can define the needed fact type: representation has context Definition: the context occurs before or around the representation and helps to discern its full meaning My previous objections to this still stand. The text "the car has a full tank of gas" has different meanings in different contexts. Example 1 (defining a fact type): [car] is fueled Definition: the car has a full tank of gas You mean: The [car] has a full tank of gas. That [car] is a variable. And in the context of a definition, the variable is bound to the universal quantifier: For all [car], [car] is fueled == [car] has a full tank of gas And the semantic formulation must convey that binding. That is, in this occurrence, the representation [car] is the sign "car" for a universal variable ranging over the domain that is the extension of the concept 'car'. Example 2 A car with license number 98765 is ready for renting. The car has a full tank of gas. This occurrence of "the car" is a syntactic shorthand for "the car with license number 98765" In its textual/spoken context, it constitutes a reference to the previous "definite reference". That is, the representation [the car] is the sign sequence "the car" for a unitary variable ranging over the set { c from 'car' such that c has license number "98765" } Here we see the same expression used in two different representations. In the first case the context is the fact type form "[car] is fueled". In the second case the context is the statement "A car with license number 98765 is ready for renting." In each case, the context helps to understand the full meaning of the representation. Not quite. Recall that a Representation is a pair: (Expression, Meaning). I agree that there is a difference in the representation associated with the Expression "the car" in the two cases. The same sign has two different meanings. But the representation itself doesn't depend on the context. What depends on the context is the interpretation -- the *decision* as to what Meaning is associated with the Expression. Once that decision is made and the Meaning is established, we have exactly one Representation -- the sign "car" has that Meaning. The remaining "context ideas" are relationships among meanings. The meaning of the first example is a fact type and the meaning of the second is a proposition. This is inaccurate. They are both propositions. The first is asserted as part of the conceptual schema; the second is asserted as part of the information base. In the first case, the context of interpretation for the Expression "the car" is the definition of a unary fact type whose domain is the extension of 'car'. It is that context that allows the interpreter to decide that the Meaning of "the car" is a reference to the universal variable. The formal definition is itself a proposition, and the "actuality of definition" asserts it. In the second case, the context of interpretation is a sequence of assertions of (temporal) propositions. The Expression "the car" appears in one such assertion, namely that of a proposition involving the unary fact-type '[car] is fueled'. "The car" therefore apparently designates a variable that is intended to be bound. The interpretation problem is to find its domain, and for that we use knowledge of natural language patterns. Note that: "It has a full tank of gas." has exactly the same meaning in the second context and relies on a slightly different natural language pattern to be interpreted as the same definite reference. In that case, we have a different Representation as well -- a different Expression with the same Meaning. Currently, SBVR gives us words to relate each of the two representations to the one text and to their respective meanings. But SBVR does not yet give us words to relate the representations to their respective contexts. Adding 'context' and 'representation has context' is what is needed. I disagree. If we have words/structures that allow us to represent the two distinct Representations, i.e. pairs of the form (Expression occurrence, Meaning), what else do we need? We might need the idea that the occurrence of an Expression has some kind of "context" which is a physical relationship to other Expression occurrences, and possibly to their Representations. But it requires an understanding of the rules of the Expression language to be able to relate that physical structure to an intepretation of the Expression. [XMI parsers and Java compilers do exactly that. And it is not a problem for Germans to understand: "Es ist nicht alles Gold, was glaenzt." ("It is not everything gold what glitters.") But to make that readily comprehensible to speakers of English, we use slightly different natural language patterns. It is those expression patterns that are used in extracting the meaning from the statement and its context, and they are language dependent.] Don says: Note that not all SBVR enabled tools will parse rules and definitions to create semantic formulations. If tools that don't are going to pass information via XMI to a tool that does, then they will need to be able to communicate context. Otherwise the tool that parses won't have what it needs to make sense out of the texts of statements and definitions. That is the practical reason for making this addition. But that means that tools that do not determine Meaning, and therefore cannot construct Representations, need to convey some context for Expressions. I would hope that SBVR XMI provides all the SBVR-defined "contexts" in which such Expressions can occur. And I'm not sure what other "contexts" could be constructed. I guess I need an example of what Don has in mind. -Ed -- Edward J. Barkmeyer Email: edbark@nist.gov National Institute of Standards & Technology Manufacturing Systems Integration Division 100 Bureau Drive, Stop 8263 Tel: +1 301-975-3528 Gaithersburg, MD 20899-8263 FAX: +1 301-975-4482 "The opinions expressed above do not reflect consensus of NIST, and have not been reviewed by any Government authority." Subject: RE: issue 9933 -- SBVR FTF issue Date: Mon, 16 Oct 2006 12:44:53 -0700 X-MS-Has-Attach: X-MS-TNEF-Correlator: Thread-Topic: issue 9933 -- SBVR FTF issue Thread-Index: AcbxVxfD1YFYBXtFRY+/fdOIHRaspwAAyW8A From: "Baisley, Donald E" To: "SBVR-FTF" X-OriginalArrivalTime: 16 Oct 2006 19:44:54.0296 (UTC) FILETIME=[8D61E980:01C6F15B] X-MIME-Autoconverted: from quoted-printable to 8bit by amethyst.omg.org id k9GJfaxh016983 Ed wrote: > I would hope that SBVR XMI provides all the SBVR-defined "contexts" > in which such Expressions can occur. SBVR does not address any of the structure of terminological works -- that is a space that ISO covers already. If a necessity statement accompanies a glossary entry for a term, SBVR XMI will communicate the term and the necessity statement but does not have a way to communicate that one accompanies the other. If a definition is given under a fact type form, and then ten other fact type forms are also given for the same fact type, SBVR XMI will communicate all of the fact type forms for the one fact type and the definition given for the fact type, but SBVR XMI will not communicate that the definition was given for a particular one of the forms. That is the whole point of the issue. A single fact type, 'representation has context', will support what is needed. Regards, Don To: sbvr-rtf@omg.org Subject: Issue 9933 - proposed resolution X-KeepSent: 42F3A3DC:308AE2F6-85257634:0075ED4F; type=4; name=$KeepSent X-Mailer: Lotus Notes Release 8.0.2 HF623 January 16, 2009 From: Mark H Linehan Date: Thu, 17 Sep 2009 17:28:48 -0400 X-MIMETrack: Serialize by Router on D01MLC04/01/M/IBM(Release 8.0.2FP1 ZOS802FP1HF3|June 25, 2009) at 09/17/2009 17:29:19 Donald asked me today to draft a proposed resolution for 9933: -------------------------------- Mark H. Linehan STSM, Model Driven Business Transformation IBM Research phone: (914) 784-7002 or IBM tieline 863-7002 internet: mlinehan@us.ibm.com Issue 9933.doc Disposition: Resolved OMG Issue No: 9933 Title: Context for understanding representations Source: Microsoft (Mr. Don Baisley, don.baisley@microsoft.com) Summary: A definition given for a fact type commonly refers to the roles of the fact type using placeholders from a form of expression of the fact type. But the SBVR vocabulary provides no relationship between a definition and the form of expression for which the definition makes sense (the one having the placeholders used in the definition). A new fact type must be added to SBVR in order to capture this relationship. Otherwise there is insufficient context to interpret the definition. The problem occurs because there can be many forms of expression for a fact type, and these can uses different designations for placeholders. A similar problem occurs where a note or example for a concept only makes sense in relation to a particular designation or form of expression, such as when the example or note refers to the designation or a placeholder of the form of expression. Similarly, the expression of a reference supporting a concept often does not include the designation used in a source document because the reference is given in relation to that particular designation for the concept. But if there are many designations for the concept, the reference cannot be adequately interpreted with respect to the source document. Also, a note about a rule might make sense only in the context of a particular statement of the rule because the note refers to specific words used in the rule. But the same rule can be represented by many different statements. Recommendation: Add a new fact type to the Meaning and Representation Vocabulary after .representation represents meaning.: representation relates to representation And explain that the first representation is understood in the context of the second representation. Resolution: No change. This is an issue of how vocabulary and guidance entries are given, not how they are captured in the SBVR metamodel. In SBVR Structured English, as described in Annex C.3.2, placeholders in different fact type forms that have the same designation correspond to the same fact type role. Hence any references to these designations unambiguously refer to specific fact type roles. Notes about a rule can qualify placeholder designations as needed to disambiguate them. The proposed fact type would not sufficiently disambiguate placeholder references in cases where a rule uses the same fact type multiple times, as in: It is obligatory that a is greater than b and b is greater than c. Disposition: Resolved