Abstract

The ISO Orange Report TR 9007 – Concepts and Terminology for the Conceptual Schema and the Information Base is a widely acclaimed early publication that recognized the importance of formalizing semantics of information:  To understand the information needs of an enterprise we need models of the enterprise, the things and affairs it must know about, and the parts of the enterprise involved in information exchanges. This column is the fourth in a series giving some of the history of the Orange Report and summarizing its basic subjects.  Last time, we learned about the basic facts covered by the Orange Report.  In this instalment the author recaps the modelling approaches that were studied as candidates for a conceptual modelling approach.

Modelling Approaches

In the late seventies, several approaches to information and data modelling were available.  Data modelling was in the first place dedicated to the form in which information was manipulated, in particular, in computer systems and data bases.  Information modelling was more inclined to concentrate on the meaning — the semantics of the information.  And as we concentrated on the conceptual level of information handling, it was felt that we should mainly consider the information modelling approaches.

The Group decided on three Categories to be studied as potential candidates for a conceptual modelling approach:

• Entity-Attribute-Relationship approaches,
• Binary and Elementary N-ary Relationship approaches,
• Interpreted Predicate Logic approaches.

The Entity-Attribute-Relationship Approaches

The Entity-Attribute-Relationship approaches clearly have their roots in the data base modelling techniques of the sixties and seventies, in particular the Codasyl approach^[CODA1978].  Well-known authorities on the subject were Bachman^[BACH1969], Chen^[CHEN1976], Olle^[OLLE1978], and Tardieu^{[TARD1976] [TARD1980]}.

Entity-Attribute-Relationship (EAR) approaches are based on the use of the following primitives:

• entities,
• relationships among entities.
• attributes, associations between values and entities, or between values and relationships.
• values.

These approaches also make use of the notions of type and occurrence applied to each of its primitives.

The origins of these approaches are the data modelling practices of the early seventies.  Originally only binary (dyadic) relationships were allowed, and attributes of relationships were not recognized.  However, more recent developments have resulted in variants that allowed n-ary relationships between entities and allowed relationships to have attributes.

The EAR approaches can be characterized as being oriented towards the definition of static aspects.  Therefore, generally speaking, they can describe only partially the various rules of the universe of discourse.  The EAR approaches often imply special kinds of propositions that are grouped together and expressed in single macro constructs.  They do not provide for explicit distinction between lexical and non-lexical entities.

The entities had a direct correspondence to the semantics of data base records; the data elements referred to attributes, and the relations between different data base records were closely reflecting basic propositions associating important entities.  Therefore, it was very easy to reflect the structural aspects of the entities and states of affairs in a universe of discourse in a data base system based on these approaches.

It was more difficult to model other dependencies between entities and states of affairs and, in particular, the dynamic aspects like change and transitions of affairs.

Binary and Elementary N-ary Relationship Approaches

Historically, the Binary Relationship (BR) approaches have their roots in certain approaches in artificial intelligence and linguistics, dealing with "semantic networks" and other similar notions.  They were introduced to the data base community in the early seventies by authors such as Abrial^[ABRI1974] and further developed by several other authors in the mid-seventies — among them Bracchi^[BRAC1976], Durchholz^[DUR1974], Kent^[KENT1978], Nijssen^{[NIJS1978] [NIJS1980]}, Senko^[SEN1977].

The BR approaches distinguish entities from entity-names.  They do not distinguish between attributes and relationships.  Only relationships that are binary are recognized.  The BR approaches are based on three primitives:

• entities,
• entity-names,
• binary relationships.

Also, these approaches make use of the notions of type and occurrence applied to each of its primitives.

The Binary Relationship approaches started out with the capability of defining mainly static aspects, but in recent years they have been extended to handle dynamic aspects as well.  These approaches are now able to describe all rules that are relevant for the universe of discourse.  Variants of these approaches distinguish explicitly between lexical and non-lexical entities.

Developments with the same roots and philosophy, also in the mid seventies, resulted in the elementary n-ary relationship approaches — e.g., Falkenberg^[FALK1979].  These approaches do not restrict an elementary proposition to be about exactly two entities, but allow description of elementary propositions involving one, two, or more entities (elementary n-ary relationships).

The basic idea of all these binary and n-ary relationship approaches is to model the universe of discourse explicitly and separately using sentences that express simple elementary propositions, thus not introducing a specific grouping of those elements.  Grouping was not considered to be at the conceptual level.

Interpreted Predicate Logic Approaches

The Interpreted Predicate Logic (IPL) approaches — as proposed by authors such as Steel^{[STEEL1978] [JARD1985] [GRIET1978]}   — perceive the universe of discourse as solely consisting of:

• entities,

for which

• propositions

hold.

The conceptual schema and information base constitute a description consisting solely of a set of sentences encoded in some formal language based on formal logic.  Such sentences are composed of:

• terms and variables
• predicates
• logical connectives
• quantifiers

The terms and variables refer to the entities in the universe of discourse, and the sentences express the propositions about those entities.

The essence of the IPL approaches is the establishment of an interpreted, axiomatized, deductive, formal system of logic describing the universe of discourse without placing any modelling constraint on the universe of discourse itself.

The basic principles of these approaches are equally well applied to both static and dynamic aspects of the universe of discourse and those of its description in the conceptual schema and information base.  Therefore the approaches are able to describe all the rules as prescribed for the universe of discourse and ipso facto its description.  They also provide for explicit distinctions between lexical and non-lexical entities.

Some variants of the IPL approaches apply a very limited set of elementary constructs that are built into the information processor, and use these to "generate" and construct the full conceptual schema and information base relevant for the chosen universe of discourse.  Others include more complex constructs and capabilities in their basic set to increase the ease and convenience of a user in expressing all kinds of propositions about the universe of discourse.

They all provide for dynamic change of the conceptual schema as well as the information base.  Several of them also provide for dynamically extending the expressive power of the language used by adding possible constructs using the capabilities that are already present.

Copyright ©2008 J J van Griethuysen

---

Next time, we will conclude our look at the Orange Report by considering the far-reaching effects it has had.  And the author will share his perspective on how the Orange Report relates to the Semantics of Business Vocabularies and Business Rules (SBVR).

References

[ABRI1974]  J.R. Abrial, "Data Semantics," Data Base Management, Proceedings IFIP TC2 conference, Cargese, 1974; North-Rolland Publishing Company.  

[BACH1969]  C.W. Bachman, "Data Structure Diagrams," Data Base 1, No.2, 1969, Special Interest Group Processing, Publication of ACM on Business Data.  

[BRAC1976]  G. Bracchi, P. Paolini, and G. Pelagatti,  "Binary Logical Associations in Data Modelling," Modelling in data base management systems, proceedings IFIP TC2 Conference,Freudenstadt 1976; North-Holland Publishing Company.  

[CHEN1976]  P.P. Chen, "The Entity-Relationship Model Toward a Unified View of Data," ACM TODS, Vol. 1, No. 1, 1976.  

[CODA1978]  Codasyl Data Base Task Group, January 1978 Report, CODASYL JOD, 1978.  

[DUR1974]  R. Durchholz and G. Richter, "Concepts for Data Base Management Systems," Data Base Management, Proceedings IFIP TC2 conference, Cargese, 1974; North-Holland Publishing Company.  

[FALK1979]  B. Breutmann, E. Falkenberg, and R. Maurer, "CSL:  A Language for Defining Conceptual Schemas," Data Base Architecture (G. Bracchi, G.M. Nijssen, eds.), North-Holland 1979, 237-256.  

[GRIET1978]  J.J. van Griethuysen, Notation Methods for an Information Model.  Internal Philips Publication, 1978.  

[JARD1985]  D.A. Jardine,  "Semantic Agreement and the Communication of Knowledge," Proceedings of the IFIP TC 2.6 Conference on Database Semantics, Hasselt, 1985, North-Holland Publishing Company, 1986.  

[KENT1978]  W. Kent, Data and Reality, North-Holland Publishing Company, 1978.  

[NIJS1978]  G.M. Nijssen, A Framework for Discussion in ISO/TC97/SC5/WG3 and Comments on 78.04/01 and 78.05/03, Working Paper for ISO/TC97/SC5/WG3, 1978.  

[NIJS1980]  G.M. Nijssen, "A Framework for Advanced Mass Storage Applications," Medinfo 80, Proceedings of the Third World Conference on Medical Informatics, North-Holland Publishing Company, 1980.  

[OLLE1978]  T.W. Olle, "Multistage Data Definition in a multicomponent DBMS architecture," Data Bases:  Improving usability and responsiveness, (B. Schneiderman, ed.), Academic Press, 1978.  

[SEN1977]  M.E. Senko, "Conceptual schema, abstract data structures, enterprise descriptions," International computing symposium 1977; North-Holland Publishing Company.  

[STEEL1978]  T.B. Steel Jr., "A Modest Proposal for a Conceptual Schema Language," Proceedings of SHARE LI, vol. 1, SHARE Inc, Chicago, 1978.  

[TARD1976]  P. Moulin, J. Randon, and H. Tardieu, "Conceptual model as a data base design tool," Modelling in data base management systems, proceedings IFIP TC2 Conference, Freudenstadt 1976; North-Holland Publishing Company.  

[TARD1980]  H. Tardieu, H. Heckenroth, D. Pascot, and D. Nanci, "A method, a formalism, and tools for data base design:  Three years of experimental practice," Entity-Relationship Approach to System and Analysis, (Chen, P.P., ed.), North-Holland Publishing Company, 1980.  

# # #