Semantische interoperabiliteit met behulp van een bedrijfsbrede taxonomie Wat kunnen we leren van IBM’s IFW Business Data Concepts Classification R. Claassens, 28-01-2005 Master Class, Information Management 19 (2004), Tias Business School, Tilburg

Doelgroep Information Management Inhoud Situatieschets Probleemschets Oplossingsrichtingen Ontologie/taxonomie Faceted Approach IBM IFW Investeren? Conclusies en advies

Doelgroep

Information Management

Inhoud

Situatieschets

Probleemschets

Oplossingsrichtingen

Ontologie/taxonomie

Faceted Approach

IBM IFW

Investeren?

Conclusies en advies

Down- load file Down- load file Down- load file Screen scrape Screen scrape Browser HTTP/XML Trans- action file Trans- action file Trans- action file Trans- action file Message queue Message queue Message queue FTP Sockets E-mail Message XML/ HTTP Gateway RPC CICS gateway APPC SMTP CICS gateway ORB Applications From Mergers and Acquisitions Legacy Applications Purchased Packages Applications in Trading Partners E-Marketplaces End-User Development Autonomous Divisions Outsourced and ASP Applications Situatieschets “ Jaguar” “ Geautomatiseerde systemen worden steeds meer samengesteld uit delen gedistribueerde functionaliteit die vaak onafhankelijk van elkaar ontwikkeld zijn, die zich op verschillende platforms bevinden en die zich in principe overal kunnen bevinden. Focus : semantische interoperabiliteit/integratie binnen het bedrijf Jaguar?

Stands for Refers to Symbolize Concept(s) (in mind) – or thougt idea intension “ Jaguar” 1) The Meaning Triangle (Odgen, Richards, 1923) 2) Example of the ambiguity of symbols Symbol(s) –or term label code Referent(s) -or thing object extension (Based on Ogden & Richards, 1923) Probleemschets before: Frege, Peirce; see (Sowa 2000)

Concept(s) (in mind)

– or

thougt

idea

intension

Symbol(s) –or

term

label

code

Referent(s) -or

thing

object

extension

... Human Agent 1 (HA1) Human Agent 2 (HA2) exchange signs, e.g. nat. language ‘‘ JAGUAR“ Internal models Formal models exchange signs, e.g. protocols MA1 HA1 HA2 MA2 a specific domain, e.g. animals Machine Agent 1 (MA1) Machine Agent 2 (MA2) (H2H) (H2A),(H2A2H),(A2H) (A2A) = flow of communication and also the flow of semantics (H2H) (H2A),(H2A2H),(A2H) (A2A) Human to human Human to application Application to human Application to application Things in the real world Concepts / Semantic structures Symbols / Syntactic structures The Meaning Triangle & (Based on Maedche, 2002) Probleemschets Stands for Refers to Symbolize Symbol Referent Concept

...

1 2 3 4 5 6 1 2 3 4 5

... Human Agent 1 (HA1) Human Agent 2 (HA2) exchange signs, e.g. nat. language ‘‘ JAGUAR“ Internal models Formal models exchange signs, e.g. protocols HA1 MA2 a specific domain, e.g. animals Machine Agent 1 (MA1) Machine Agent 2 (MA2) Ontology Description Ontology Formal Semantics =The ontological commitment refers to agreements on the use of the shared vocabulary by the agents committed to the ontology Things in the real world Concepts / Semantic structures Symbols / Syntactic structures The Meaning Triangle commit MA1 commit commit commit HA2 commit (Based on Maedche, 2002) Oplossingsrichting : ontology Stands for Refers to Symbolize Symbol Referent Concept

...

a philosophical discipline The Science of Being Aristotle, Metaphysics. Immanuel Kant, Charles Sanders Pierce Ontology Comes from Borrowed by Used in Ontology (Based on Guarino, 1998) Research fields Knowledge engineering Knowledge representation Qualitative modeling Language engineering Information modeling Information integration Knowledge management and organization Database design Agent based system design Information systems Enterprise integration Natural language translation Medicine Mechanical engineering Standardization of product knowledge Electronic commerce Geographic information systems Legal information systems Biological information systems

a philosophical discipline

The Science of Being

Aristotle, Metaphysics.

Immanuel Kant, Charles Sanders Pierce

Stands for Refers to Symbolize Symbol Referent Concept 1)The Meaning Triangle Ontology Philosophical viewpoint Computer science viewpoint An ontology is an explicit specification of a shared conceptualisation (Gruber 1995) An ontology is a shared understanding of some domain of interest. (Uschold et al. 1996) ... philosophical discipline, branch of philosophy that deals with the nature and the organisation of reality.” (Guarino 1998) 2) Utmost views on ontology

philoso-phical ontology pragmatic ontology top level ontology universal ontology domain specific ontology general ontology task specific ontology task inde-pendant ontology language inde-pendant ontology language inde- pendant ontology formal ontology not formal onto-logy VIEW specific ontology LEVEL SUBJECT PURPOSE LANGUAGE FORMALIZING application specific ontology Guarino, Nicola (1998). Formal Ontology and Information Systems,. In: Formal Ontology in Information Systems, Proceedings of the First International Conference (FOIS'98) , June 6-8, Trento, Italy, 3-15. Ed. Nicola Guarino. Amsterdam: IOS Press. Bodil Nistrup Madsen (2002), based on a.o.: ontology Ontology

Ontologieën: een spectrum Catalog Terms/glossary Thesauri (“narrower term” relation) Informal “is-a” hierarchy Formal “is-a” hierarchy Formal instance Frames (properties) Value restrictions Disjointness, inverse, equivalence, etc. General logical constraints Ontologies Dewey Decimal System Data dictionary Yahoo directory Biological taxonomy, RDB schema, OODB schema XML schema CLOS, C++, Java, RDF UML DAML+ OIL (OWL) DAML-L (rules), KIF Text search engine thesauri Taxonomies Expressiveness vlgns Deborah McGuinness *based on AAAI ’99 Ontologies panel – McGuinness, Welty, Ushold, Gruninger, Lehmann

Ontologieën: bedrijfbrede interoperatibility Reduce complexity (thru generalisation and simplicity) Simpler models can have greater functionality. Additional functionality does NOT have to lead to additional complexity. Reduced life cycle costs Meet currently unspecified/unknown requirements (thru fruitfulness) Reduced time to market Functionality/scoop Levels Of generality Narrower Wider Lower Higher “ Traditional Modeling” “ Ontology Based Modeling” ( Partridge , 1998) Goals

Reduce complexity (thru generalisation and simplicity)

Simpler models can have greater functionality.

Additional functionality does NOT have to lead to additional complexity.

Reduced life cycle costs

Meet currently unspecified/unknown requirements (thru fruitfulness)

Reduced time to market

English Prose Poetry Sonnet Ballad German Prose Poetry Sonnet Ballad French Prose Poetry Sonnet Ballad To illustrate the differences between faceted and enumerative classifications, consider the class Literature. In an enumerative scheme we might have the following subclasses: Language facet : English German French Form facet : Prose Poetry Sonnet Ballad In a faceted scheme we might have the following instructions and subclasses: Language facet is cited before form facet Citation order: Faceted classification a Library discipline Ranganathan 1939

Postulated Ontology Synthesized Clusters Ontology is modified based on how it maps to discovered clusters x s t u v w D F E B C A Clusters are mapped to ontology 2) Bottom-up 3) Revising & validating 1) Top-down R. Prieto-D í az (2002) a faceted Approach to building Ontology's

The Information Framework Three views Three levels Ten columns Five rows Fifty cells (Modelware International, 1999) III) Technical View II) Business View I) Organisation View Structure Skills Strategy Data Functions Workflow Solutions Interface Networks Platform Domain Concept (A-level) Domain Classification (B-level) Generic Template (C-level) Design Context (C’-level) Operational Bound (D-level) Deconstruction level Composition level Implementation level Types of infomation Levels of constraint Business Data Concepts Classification =

Three views

Three levels

Ten columns

Five rows

Fifty cells

IFW Framework -3 Layers of the data column Deconstruction level Composition level Implementation level Conceptual Logical Physical A-level B-level C-level & C’-level D-level 9 data concepts 27 classification hierarchies 54 business objects (Modelware International, 1999) Involved Party (IP) Arrangement (AR) Conditions (CN) Product (PD) Location (LO) Classification (CL) Business Direction Item (BD) Event (EV) Resource Item (RI)

Involved Party(IP) INVOLVED PARTY TYPE Individual INDIVIDUAL GENDER Female Male Organization ORGANIZATION LEGAL STRUCTURE TYPE Corporation Partnership IP Descriptor IP DESCRIPTOR TYPE IPName component IP NAME COMPONENT TYPE Given Name Name Initial Family name IP Relationship IP RELATIONSHIP TYPE IP/IP-relationship IP/IP RELATIONSHIP TYPE IP is spouse of IP IP is employee of IP IP is customer of IP INDIVIDUAL EMPLOYMENT STATUS Working Individual Not Employed Individual 1. Fundamental hierarchy 2. Descriptive hierarchy 3. Relationship hierarchy Answer(s) Question = + scheme Explanation of the B-level - Concept Involved party (IP) (Modelware International, 1999)

The Information Framework Three views Three levels Ten columns Five rows Fifty cells (Modelware International, 1999) III) Technical View II) Business View I) Organisation View Structure Skills Strategy Data Functions Workflow Solutions Interface Networks Platform Domain Concept (A-level) Domain Classification (B-level) Generic Template (C-level) Design Context (C’-level) Operational Bound (D-level) Deconstruction level Composition level Implementation level ORM ORM ORM ORM ORM ORM DM DM DM OM DW OM DW FM FM WM OM OM WM WM WM OM OM OM Types of infomation Levels of constraint DM = Financial Services Data Model ORM = Financial Services Organization Model DW = Data Warehouse WM = Financial Services Workflow Model FM = Financial Services Function Model OM = Financial Services Object Model Content Models:

Three views

Three levels

Ten columns

Five rows

Fifty cells

Management Objectives - for the Information Technology Portfolio Infor- mational Strategic Transactional Infrastructure Increased control Better information Better integration Improved quality Business integration Flexibility & agility Reduced marginal cost of IT Reduced IT costs over time Standardization Increased sales Competitive advantage Competitive necessity Market positioning Innovative services Cut costs Increased , throughput Positionering van de investering in taxonomieën -Information investment portfolio- Based on (Weill, P. & Broadbent, M. 1998)

Management Objectives - for the Information Technology Portfolio

Increased control

Better information

Better integration

Improved quality

Business integration

Flexibility & agility

Reduced marginal cost of IT

Reduced IT costs over time

Standardization

Increased sales

Competitive advantage

Competitive necessity

Market positioning

Innovative services

Cut costs

Increased ,

throughput

Direct voordelen ondersteuning bij het oplossen van communicatieproblemen tussen verschillende expertisedomeinen; vorm en structuur waardoor redundantie in concepten en bijbehorende termen eenvoudig kan worden voorkomen; uitbreidbaarheid zonder dat de hoofdstructuur van de taxonomie gewijzigd hoeft te worden; een stabiel uitgangspunt voor meer formele modellen, die noodzakelijk zijn voor volledige geautomatiseerde communicatie en verwerking; een mogelijke basis voor een kennisapplicatie waar gebruikers geïnformeerd worden over de betekenis van de termen en hun afhankelijkheden, die binnen een bedrijf worden gehanteerd. Indirecte voordelen Het sneller en op een semantisch verantwoorde wijze kunnen koppelen van heterogene en geautomatiseerde systemen. Minder fouten en conflicten als gevolg van minder onduidelijkheden over de semantiek van gegevens. Conclusies IFW Business Data Concept Classification

Direct voordelen

ondersteuning bij het oplossen van communicatieproblemen tussen verschillende expertisedomeinen;

vorm en structuur waardoor redundantie in concepten en bijbehorende termen eenvoudig kan worden voorkomen;

uitbreidbaarheid zonder dat de hoofdstructuur van de taxonomie gewijzigd hoeft te worden;

een stabiel uitgangspunt voor meer formele modellen, die noodzakelijk zijn voor volledige geautomatiseerde communicatie en verwerking;

een mogelijke basis voor een kennisapplicatie waar gebruikers geïnformeerd worden over de betekenis van de termen en hun afhankelijkheden, die binnen een bedrijf worden gehanteerd.

Indirecte voordelen

Het sneller en op een semantisch verantwoorde wijze kunnen koppelen van heterogene en geautomatiseerde systemen.

Minder fouten en conflicten als gevolg van minder onduidelijkheden over de semantiek van gegevens.

Advies IFW Business Data Concept Classification Mocht een bedrijf ervoor kiezen om een bedrijfsbrede taxonomie te gaan opzetten, dan is het advies om de “best practices” van IBM’s IFW Business Data Concepts Classification in ieder geval te bestuderen en eventueel over te nemen.