The ORM Foundation

On December 11, 2007 OMG declared SBVR a standard. A well selected subset of SBVR corresponds to CogNIAM (the latest version of NIAM with Cognition enhanced, hence concept definitions and fact type forms are first class citizens) or ORM. For persons interested in introductory descriptions of SBVR and how it relates to CogNIAM, go to www.brcommunity.com and register. From October 2007 I have a monthly column how CogNIAM and SBVR relate. The official standard is not the most readable document of this century; I am glad to say that Annex L is a description of how the 2007 version of NIAM relates to SBVR. Terry has written two other annexes about ORM.

SBVR is definitely a fact based approach in my view and you probably agree when you read  Clause 10 of the specs. I am glad to say SBVR adds some very useful extensions.

Please let me know your questions and I will try to answer.

Kind regards

Sjir Nijssen 

For people interested in ORM, CogNIAM and SBVR, it is good to know that the SBVR Foundation holds a three day tutorial-conference June 9-11, Utrecht, The Netherlands. Go to the website www.sbvrfoundation.eu and you see that Terry gives a full day tutorial on logic as the base for SBVR. This is a unique opportunity to learn from the best expert in this area.

 Regards

 Sjir Nijssen

I was asked: how can you say SBVR is a fact based approach?

Well, first let me stress that SBVR adds very valuable functionalities to fact orientation proper.

The fact based approach of SBVR is clear from Clause 10 and other parts of the spec.

Look e.g in Clause 8, 8.5

There is only one fact type expressing what the contents of a fact model is:

fact model includes fact 

Regards

Sjir Nijssen

Can we assume that CogNIAM, ORM and SBVR are 3 representations of business knowledge that can be translated to each other without any loss?

It is possible when a certain set of rules or business practice is followed to transform automatically between SBVR, CogNIAM and ORM. Hence a certain form of Structured English, CogNIAM and ORM can be considered as SBVR compliant languages. Hence we can speak of three speech communities belonging to one semantic community. This aspect of SBVR gives a new degree of freedom not seen before in conceptual modeling.  

The rules will be published in the near future.

Regards

Sjir Nijssen

My recommendation to the fact orientation community is to adopt many of the SBVR functionalities that are not available in CogNIAM or ORM.

In CogNIAM we have last year added several SBVR functionalities. One of the sets of useful functionalities is around communities, both semantic and speech, bodies of shared meanings and bodies of shared concepts. In makes communication about conceptual modeling in large organisations more manageable.

A semantic community is a community "whose unifying characteristic is a shared understanding of the things they have to deal with". SBVR, Clause 11.1.1.1.

A speech community is a "subcommittee of a given semantic community whose unifying characteristic is the vocabulary and language that it uses".

Body of shared meaning is " a set of concepts and elements of guidance for which there is a shared understanding in a given semantic community."

Body of shared concepts is "all of the concepts within a body of shared meanings."

Hi Roel,

No - I don't think that as things stand, CogNIAM, SBVR and ORM2 (or ORM1) can be "translated into each other"
ORM1 is a subset of ORM2 so that isn't much of a problem.
However, to the best of my knowledge, There are many differences between SBVR and ORM that need to be resolved before any "translation" could occur.
I don't know much about CogNIAM so I can't comment here.

Ken

Hi Ken,

 I'm not very familiar with ORM yet, but since 1989 busy with NIAM, UI and perFact. I suppose ORM is not that different is it's foundation. So I expect that any SBVR construct can be represented in some way in SBVR. I can imagine that ORM does not have a symbol for all of the present constructs of SBVR. The above representations don't have it either, and I believe even CogNIAM does not yet have a symbol for each SBVR construct. Would it be interesting enough to add for instance a symbol for "communities" (see posting of Sjir) to ORM?

Hi Roel,

I don't know anything about UI and perFact so I can't comment.
However, as I see it, although superficially similar, NIAM and ORM have a significant difference.
For me, the crucial difference is that ORM meets the criteria required of a formal language. (See Terry Halpin's 1989 PhD thesis, referenced in the Library). Indeed, the InfoModeler tool of the 1990's came with a document that described "FORML" = Formal Object-Role Modeling Language.

Regarding SBVR, I may misunderstand something but right now I don't understand the logic in suggesting that ORM should have a symbol for SBVR constructs. ORM is based on set theory. Each "object" in an object-role model represents a domain and a set of permissible values.
Thus, it seems to me to be quite unnecessary to have a special symbol for "Communities".

With ORM, everything is already there to be able to define an object-role model about "communities" as mentioned by Sjir and defined in SBVR Clause 11.1.1.1

As an example, I have created a sample object-role model of SBVR Figure 11.1.1 and uploaded it to the Library>Standards>SBVR.
The object-role model UoD is not quite the same as the SBVR diagram because I felt that diagram 11.1.1 omits certain aspects of the concept "community" that it purports to show. Specifically, I felt that if you really want to define "meaning" then surely you have to define structures for all of the senses.

I used VEA 2003 for this and it took me about 15 minutes.

Ken

ORM, CogNIAM and SBVR are all fact-oriented approaches, and it is possible to define a common core metamodel based on an intersection of these approaches, and hence enable translation between useful subsets of them. ORM and CogNIAM have much in common, and the difference between them is not so much one of formality (as both can be given a logical foundation) but rather aspects such as atomicity of fact types and constraint coverage. Work has already begun on proposing a common metamodel to facilitate model exchange between ORM and CogNIAM.

The NORMA tool for ORM 2 is designed for extensibility. The notion of semantic community could be a useful extension to ORM. While interesting, Ken's model on communities seems to have misunderstood the SBVR intent for this concept (e.g. symbols are not communities).

Terry,

Of course I agree that "symbols are not communities". (The label is not the thing!)
However, I was interpreting the SBVR standard from the semiotic perspective.
The SBVR standard lists the term "semantic community" and gives the following definition:
      "(a) community whose unifying characteristic is a shared understanding (perception) of the things that they have to deal with"

For me, the term "word" is just a type of "symbol" so to me the term "semantic community" is just another symbol.  
Furthermore, Frege's second fundamental principle is "never ask the meaning of a word in isolation, but only in the context of a proposition."
Thus, the term "semantic community" is a symbol that has a specific meaning within the propositions used in the SBVR standard (which is an example of a UoD)

Now turning to the definition of "semantic community", my perspective on an object-role model is that it represents "a precise definition of the things in a specific UoD".
Furthermore, if a group of people agree on the content of a specific UoD, then it can be said that they have a "shared understanding".
Thus, I can rephrase the SBVR definition of a "semantic community" as "a UoD whose meaning is agreed by a group of people."
(When I got this far, I thought, ho-hum, I can't see what the SBVR concept does that cannot be done by a standard object-role model and in a more succinct way.)

Now an object-role model is a set of symbols. (graphics plus words).
And excluding the other senses, we use the senses of sight and sound to communicate meaning by means of symbols.(defining a verbal utterance also as a symbol)

This is why, when defining the model that I posted, I chose to use the term "symbol" because it seemed to me to be more efficient.
For example, because my model specifies that symbols, sounds and smells have meaning, it serves to dispense with the need for a separate definition for a "speech community". Of course you could argue that there are some groups of people (communities) who only use speech and non-word symbols to communicate.

Lastly, the SBVR definition of the term "speech community" includes the term "vocabulary" whereas the SBVR definition of the term "semantic community" does not mention the term "vocabulary". Now I may not have fully grasped something here but it seems to me that until we get telepathy going then the concept of "shared perceptions" is meaningless because as far as I know, the only way that we humans can "share our perceptions" is by using propositions that are expressed in symbols which by definition must have a form of vocabulary.   

So, I hope I have given some insight into my semiotic-based perception of the relationship between ORM and SBVR.
 

Ken

Terry,

Thanks very much for your lucid description of ORM, CogNIAM and SBVR.

I agree with you that a common core model is a very worthwile effort for the fact orientation community. I like in particular your words "useful subsets of them".

I know quite a few fact orientation users who welcome such an approach.

I assume that both of us would welcome the opinion of other fact orientation users on this endeavour. 

Regards

Sjir

Ken,

A semantic community is a group of people. Their body of shared meanings is what corresponds to a Universe of Discourse. People can join the community, or stop believing and leave the community, without changing the body of shared meanings.

A speech community is a part of a semantic community – the people in the semantic community who share a vocabulary. For example, when working with a Swiss company recently, I had to deal with three speech communities – German-speaking, French-speaking and English-speaking. Their body of shared meanings is the same, but their vocabularies are different.

Most of the semantic communities I work with are (or are centred on) enterprises – businesses, not-for-profit organizations, public sector bodies, etc. Some have only a single natural language speech community (although some do distinguish between, for example, English and American). They usually have more specialized speech communities within any natural language speech community.

For example, it’s common to find an internal vocabulary within an enterprise, with specialized terminology and jargon, and a more readily-understandable vocabulary for communicating with others in the semantic community, such as customers. This supports two speech communities – employees and employees-plus-customers.

This kind of thing also happens within enterprises.  Specialists, such as accountants or lawyers or engineers, usually have one vocabulary for talking among themselves and another for talking to the rest of the enterprise.

One of the drivers for developing SBVR was to separate concepts from their expression, and support the different ways people can – and do - talk about the same body of shared knowledge, both across natural languages and within a given natural language. An important effect is that speech communities within a semantic community could have non-overlapping vocabularies, but still understand and use the same concepts and business rules.  

Of course, to support multiple speech communities within a semantic community, you need people who understand multiple vocabularies – people who can speak both French and German, lawyers and engineers who can write in everyday natural language – but how you find them is outside SBVR’s scope. And if you can’t find them then, yes, you will need telepathy.

Regards,

John 

Hi John,

Thanks for your explanation.
My response follows.

Cheers
Ken

A semantic community is a group of people.
Thanks for the clarification. Mea culpa! In my haste to answer the post, I had assumed that the definition of each term was self contained. I can now see that there is a structuring of terms. 

·         self contained·         includes a set of named objects (object= a domain = a domain name plus a definition plus a defined set of values)

For example the object named “Country” could be defined in terms of ISO 3166 and the set of permissible values could be defined as “the country trigraphs in the standard” ·         uses predicates (facts) to link the objects.·         Includes methods to constrain the meaning of a set of propositions. (facts)

I hold the view that words derive their meaning from the context in which they are uttered. For example, to paraphrase Frege (1980(1884): x) “Never ask for the meaning of a word in isolation, but only in the context of a proposition.” 

 “A story has it that Schubert, after playing a new piano piece, was asked by a woman what it meant. Schubert said nothing but, in answer, returned to the piano and played the same music again.”

References:

Cobley, P Jansz, L (2004) Introducing Semiotics. Royston: Icon Books Ltd.

Crow, D. (2003) An Introduction to Semiotics. Lausanne: AVA Publishing SA.

Frege, G. (1980(1884)) The Foundations of Arithmetic. Evanston, Illinois: Northwestern University Press.

Ken,

The published specification of SBVR is dtc/08-01-02. That's the one you should be reading, although the definition of ‘body of shared meanings’ hasn’t changed much since dtc/06-03-02. It is now: “set of concepts and elements of guidance for which there is a shared understanding in a given semantic community”

A body of shared meanings in SBVR is "more than just a structured set of terms with definitions". In SBVR, 'concept’ includes both ‘noun concept’ and ‘fact type’ (synonym ‘verb concept’). Noun concepts play the roles in fact types. Verb phrases in fact types have meaning only in relation to the roles they connect (e.g. "runs" has different meanings in "person runs race",  "person runs company", "blood runs cold" and "crowd runs riot"). The 'elements of guidance' are the business policies, business rules, authorizations, permissions, etc. that apply to the fact types.

‘designation context’ (itself an SBVR concept) provides context where it is needed explicitly, e.g. for disambiguation of homonyms. 

The definition of ‘vocabulary’ is: “set of designations and fact type forms primarily drawn from a single language to express concepts within a body of shared meanings”.

The definition of ‘business vocabulary’ is: “vocabulary that is under business jurisdiction” . A speech community within a business decides what terms, definitions, and fact ype forms it wants in its business vocabulary - but they must be consistent with the concepts (noun concepts and fact types) in the body of shared meanings for the business. 

In the Swiss company the semantic community is centred on the business. The body of shared meanings of the people there is not everything they know – just what they have decided is needed to run their business.  Individuals will be in other semantic communities, although there probably won’t be many in both the company’s English speech community and the yodelling semantic community, nor in both the company’s German speech community and the cricket semantic community.

But within the company, if they don’t agree on meanings, and don’t create consistent definitions in all three languages, then they can’t run their business consistently. [Note that this is not about three communities in three different places, with all the complications of localization for regulation, market, work practices, culture etc. It’s about people in the same place with different first languages and, for some, with their first language as their only language.]

As for separation of meaning from expression, it’s not that hard to visualize. Think of an empty structure connecting together all the noun concepts, fact types and elements of guidance – empty in the sense of having lots of “role holes” but no text. Then think of each speech community having a set of text objects (or equivalents, such as sinographs) that would fill those role holes. Filling the role holes with a speech community’s text objects would provide the content of that speech community’s business vocabulary. You then need a transformation for readability (verbs embedded in English, at the end in German, etc), which doesn’t have to be perfect, just understandable (although no doubt people would improve it over time).

Your Schubert example is a nice story, but irrelevant. SBVR makes no claim to separate concept from meaning – in SBVR ‘concept’ is defined as a specialization of ‘meaning’. SBVR separates meaning from expression. If the lady had said “Play it again, Franz”, in Arabic or Chinese, Schubert might not have understood what she wanted, but that wouldn’t have changed the meaning.

I don’t think your armband example holds up, either. If somebody wears a black armband, it’s a black armband whatever language you say it in. If there is a fact type "person wears mourning armband" then (if it's relevant) the definitions - in English and Chinese -  would say that a mourning armband in England is black, and in China is white. Then nobody would need to be embarrassed. The  concept is  "mourning armband", not "black armband"  or "white armband".

 Regards,

John 

\s