JB June 2005

A Historical Perspective on Conceptual Modelling:

from Information Algebra to Enterprise Modelling and Ontologies

Janis A. Bubenko jrRoyal Institute of Technology, Stockholm, Sweden

janis@dsv.su.se

JB June 2005

Young and Kent (1958)“Abstract Formulation of Data Processing Problems”

•Information set/item•Defining relationship•Producing relationship•Conditions•Temporal aspects

JB June 2005

Why the need for an abstract formalism?

• Since we may be called upon to evaluate different computers or to find alternative ways of organizing current systems it is necessary to have some means of precisely stating a data processing problem independentaly of mechanization *).

*) Young and Kent, Journal of Industrial Engineering, Nov. – Dec. 1958, pp. 471-479

JB June 2005

Why Conceptual Modelling in

Information Systems work?

- to contribute to the acquisition and description of knowledge needed in the development and maintenance of information and software systems which will become, or are, active components of real world infrastructures.

JB June 2005

Modelling during four decades

Pioneeringwork -concepts

Refinement,models andextensions

The searchfor a commonframework

Participationandunderstanding

60-ties

70-ties

80-ties

90-ties

JB June 2005

Pioneers in IS modelling: 1959-70

Young and Kent 1959

CODASYL: Information Algebra 1963

"The Scandinavian School" Langefors 1965: Theoretical Analysis of Inf.Systems USA: D Teichroew, J. Nunamaker: PSL/PSA and optimisation of Information ProcessingSystems

JB June 2005

CODASYL Development Committee:

An Information Algebra (1962)The goal of this work is to arrive at a proper structure for a machine-independent problem-defining languageat the systems level of data processing. … It should help the information processing communityto clarify, understand the fundamental and essentialfeatures of data processing considerations.…With current programming languages the problemdefinition is buried in the rigid structure of an algorithmicstatement of the solution, and such a statement cannotreadily be manipulated.

Source: CACM, Vol.5, No. 4, April 1962, pp. 190 - 204

JB June 2005

Information Algebra, basic concepts

• Entity (e)• Property (q)• Property value (v)• Property value set (V)• Coordinate set (Q) e.g. Q = (q1, q2, q3)• Property space (P) of a coordinate set (Q) e.g. P=V1 x

V2 x V3• Datum point of P: d = (a1, a2, a3)• Line, Area, Glump, ….

Every entity has exactly one datum point in a property space.

A discriminatory property space for a set of entities no datum pointrepresents more than one entity.

JB June 2005

The Scandinavian School: Langefors

e = <s, a, v, t>

s system point a attribute v value t time

e = <s, a, v, t>

s system point a attribute v value t time

Langefors, 1963

* the infological and the datalogical realms

* the “elementary message”

* the “elementary file”

JB June 2005

Langefors 1966

JB June 2005

Langefors 1966 (cont)

JB June 2005

THE PERIOD: 1970-80”REFINEMENT AND

EXTENSIONS"• The 1975 ANSI/X3/SPARC (Standards Planning and Requirements Committee) report: the three schema approach

• IFIP WG 2.6 series: "Modelling in Database Management Systems” (1974)

• IFIP TC 8 on Information Systems (1976)

JB June 2005

IFIP Technical Committee 2 on Software: Theory and Practice

Working Group 2.6 on Database (started 1974, revised later) - Started the IFIP WG 2.6 conference series:

"Modeling in Database Management Systems”1974 Cargese, Corsica1975 Wepion, France1976 Freudenstadt, Germany….etc.

Abrial, Adiba, Benci, Bracchi, Codd, Date, ….Delobel, Gardarin, Falkenberg, ….Langefors, Neuhold, Nijssen, Olle, …..Senko, Spaccapietra, Sundgren, Tsichritzis, Wiederhold,...

JB June 2005

Jean-Raymond Abrial: ”Data Semantics”(1974)

Influenced by: GDBMS, Codd’s Relational Model, AI-techniques, …

Binary model

sexp

person

spouse/spouse

children/parents

sex/personofsexage/personofage

R4=rel(person, person, parents = afn(2,2), children = afn(0, ∞))

-Schema: fact types, rules-Rules: constraints, derivation rules-Internal vs external names

number

JB June 2005

A sample NIAM schema (Nijssen)

* Source: Terry Halpin, Object-Role Modeling (ORM/NIAM)

*

JB June 2005

Sample DIAM schemas

(Senko, around 1975)

JB June 2005

CADIS**:The associative data model

based on LEAP (1969)*a

bc

rp

q

x

y

w

<a,r,b><a,p,c><r,x,q>etc.

•"An ALGOL-based Associative Language", J.A. Feldman et al, CACM 12(8):439-449 (Aug. 1969). •** J.A. Bubenko jr, O.Källhammar, CADIS: Computer Aided Design of Information Systems, in Bubenko, Langefors, Sölvberg (Eds.) Computer-Aided Information Systems Analysis and Design, Studentlitteratur, 1971.

JB June 2005

Modelling research issues in the eighties

• improving the expressive power of semantic data models and adding the temporal dimension

• ”semantic modelling” vs relational data modelling

• what are we modelling? The DB? The IS?, the real world? …?

• the operational vs the deductive & temporal approach

JB June 2005

The Deductive Temporal Approach*

UoD at time t

UoD at time t+1

event

CS(t)

IB(t)

CS(t+1)

IB(t+1)

operation

partly represented by

partly represented by

induces

newart(sofa32, 5, 2500, furniture, t238).newprice(sofa32, 2700, t419).

delart(sofa32, t726).

Part of IB(t)

Picture accordingTo Antoni Olivé ”A Comparison of the Operational and Deductive Approaches to Information Systems Modeling”IFIP Congress, 1986

* Bubenko, around 1977

JB June 2005

Derivation rulesqoh(A,Q,T):- article(A,T),

newart(A,Q0,_,_,T0), T0<T,not (newart(A,_,_,_,T1), T1>T0, T1<T),findall(S, (sales(A,S,TS), TS>T0, TS<T), SLIST),findall(L, (delivery(A,L,TL), TL>T0, TL<T), DLIST),sum(SLIST, SS),sum(DLIST, DS),

Q is Q0 + DS - SS.

Constraintsincons(c9,A,T):- newprice(A,P,T), price(A,P2,T-1), P<P2

CS(t) also includes:

JB June 2005

Multi-temporal Models

Proposition: { P(a,b,c,d, …), tv, te, tt}tv = valid timete= event timett= transaction time

JB June 2005

MODELLING IN THE EIGHTIES (cont):

ISO TC97/SCS/WG3 Concepts and Terminology for the Conceptual Schema and the Information Base, Preliminary Report, 1981

Deductive and multi-temporal models, O-O models, SDM++

IFIP WG 8.1 CRIS: Comparative Review of Information System Design Methodologies conference series

CASE-tools, Design and Analysis Assistants, etc

Synergy-workshops: PL+AI+DB+SE+IS+....

TC8, WG 8.1: The FRISCO (Framework of Information System Concepts) effort, established 1988. A reportwas presented in 1996.

- the search for a common framework

JB June 2005

ISO TC97/SCS/WG3 Concepts and Terminology for the Conceptual Schema

and the Information Base, Preliminary Report, 1981

edited by J.J. van Griethuysen et al.

• Assumes the ANSI/SPARC three-schema approach• Ambitions:

- to define concepts for conceptual schema languages

- to develop a methodology for assessing proposals for conceptual schema languages

- to assess candidate proposals for conceptual schema languages

- etc.

JB June 2005

Describing the Universe of Discourse

Universe of DiscourseUniverse of Discourse Description

1

2 Representationof the abstraction system

3: Representationof the object system

Abstraction System

ObjectSystem

ConceptualSchema

InformationBase

1: Classification, abstraction, generalization, establishing rules, ….

JB June 2005

ISO TC97/SCS/WG3 Concepts and Terminology for the Conceptual

Schema and the Information Base, Preliminary Report, 1981

General notions and principles Four ”conceptual schema language

candidates” analyzed using an example Universe of Discourse The Entity-Attribute-Relationship approaches The Entity-Relationship approaches The Binary Relationship approaches The Interpreted Predicate Logic approaches

JB June 2005

Ambitions of the eighties:• to better understand and improve parts of existing methods and tools

• to harmonise different notions and methods

• to enhance the requirements capture and validation stage of the systems life-cycle

• to provide computerised assistance to the process of developing a specification

• to pay attention to human, cognitive, linguistic, and social aspects of IS

JB June 2005

On business rules

Many business rules are deeply imbeddedin programs of a company’s information system

Many business rules are deeply imbeddedin programs of a company’s information system

Rule A: If employee x has salary y and if y is greater thanz then employee x is also a manager

Rule B: All managers work full time

Rule A: If employee x has salary y and if y is greater thanz then employee x is also a manager

Rule B: All managers work full time

Vx,y (employee(x) & salary(x,y) & y > z --> manager(x))Vx manager(x) --> worksfulltime(x)Vx,y (employee(x) & salary(x,y) & y > z --> manager(x))Vx manager(x) --> worksfulltime(x)

JB June 2005

Modelling in the nineties:focus on organisational aspects, participation and understanding

… "the understanding and support of i) human activities at all levels in an organisation, ii) change, be it of the product, of the process or of the organisation, and iii) complex user organisations, and individual users" (ESPRIT 91)

JB June 2005

The nineties: Widening the scopeThe nineties: Widening the scope

Interoperable systemsSemantic heterogeneityNon-functional requirementsBusiness modelling/engineeringModelling of intentions and actorsParticipative modelling”Method knowledge” *)”Patterns”

*) e.g. the EMMSAD (Evaluation of Modelling Methods in SystemsAnalysis and Design) workshop series, start 1996.

JB June 2005

Enterprise Modelling with EKD - integrated descriptions

Goals, problems , opportunities, threats, weaknesses, constraints

Informationconcepts

(conceptualmodel)

Businessrules

Businessprocesses(control

and flows)

Actorsand

resources

Technical IS components and requirements

JB June 2005

Sample of an

Enterprise Model (EKD)

instance

To provide advancedservices to library

customers

Goal 1To minimise

library's operationalcosts

Goal 2

Deliver itemselectronically

Goal 3

High stockavailability

Goal 4

Copyright andownership of

electronic material

Problem1

Advancedcommunication and

informationtechnology

Opportunity 1supports supports

supports

hinders

hinders

Requests for electronicmaterial must be satisfied

within 3 days

Rule 1

supports

ElectronicService assistant

Role 2

Librarian

Role 1

is_respon-sible_for

Library item

Entity1

Magazine

Entity2

Information

Entity3

Book

Entity4

refers_to

Management ofelectronic

information

Process1

Customers

Ext.Process1

requests forelectronic information

responses to requests forelectronic info.

performs

The Library InformationManagement System

The superintelligent

informationlocator

To have a high servicerate to requests for

electronic information

IS Goal11

supports

To be able to locaterequested informationin 99% of all requests

IS Requirement1concerns

supports

motivates

Part of an ObjectivesModel (OM)

Part of a Business RuleModel (BRM)

Part of an Actorsand ResourcesModel (ARM)

Part of a BusinessProcess Model (BPM)

Part of anInformationModel (IM)

Part of a Technical Components and Requirements Model(TCRM)

JB June 2005

Enterprise Modelling• Purpose of modelling: not only IS design

• Models not only “what” but also “why”

• Integrates conceptual and process models of the business with objectives, actors, business rules and information system requirements

• Makes information system solutions traceable to objectives

• Makes conceptual modelling a “participatory” activity

JB June 2005

Iterative development of knowledge and models

Objectives

InformationConceptsProcesses

Actors

IS requirements

Business Rules

ConceptualModels

JB June 2005

Participation in modelling

JB June 2005

Modelling during four+ decades

Pioneeringwork -concepts

Refinement,models andextensions

The searchfor a commonframework

Participationandunderstanding

60-ties

70-ties

80-ties

90-ties

2005

- Extended scope-Standardisationefforts

Databasemodels

InformationSystem models

Modelling of ”why”, Enterprise models

Temporalaspects

User educationand participation

Domain Specific ”Ontological Models”and languages

Business rulemodelling

Formalityvs informality