Topic Maps and the Ontological World
Dr. H. Holger RathDirector Research & Developmentempolis [email protected] – http://www.empolis.com
Roadmap
n Act I: Introduction
n Scene I: What are Topic Maps?
n Scene II: How do TMs work?
n Scene III: The Family of TM Standards
n Scene IV: TMs and Related Paradigms
Roadmap cont‘d
n Act II: Allegro
n Scene V: Ontologies, Schemas, Templates
n Scene VI: Class hierarchies
n Scene VII: Inferencing
n Scene VIII: Consistency constraints
n Scene IX: Topic Map Query Language (TMQL)
n Epilogue: Conclusions
Act I:
Introduction
Scene I:
What are Topic Maps?
An overview ...
The Sound-bites
n “GPS of the information universe”
n “A new paradigm for organizing, maintaining, and navigating information”
n “The bridge between Information Management and Knowledge Management”
Topic Maps are ...
n Standardized:
n An ISO standard describing knowledge structures, electronic indices, classification schemes, ...
n Web enabled:
n XML Topic Maps (XTM) are ready to use
n Designed to:
n manage the info glut
n build valuable information networks above any kind of resources / data objects
n enable the structuring of unstructured information
The 3rd Prophecy
Unified
Universal
Semantic
HTTP:// XML TopicMapsTopicMaps?! ?!
19911. Revolution
19972. Revolution
200320033. Revolution3. Revolution
By Tim Berners-Lee (father of the Internet)
Topic Maps – A Promising Technology
nMetadatan Topic Map data is not part of the info assets
n SearchnSearch in more precise topics and not in full text
n Linkingn TMs are well-organized link networks
n Knowledge structuresn TMs are a base technology for knowledge
representation
Scene II:
How do TMs Work?
Brief intro ...
Example: Back-of-the-Book Index
Don Giovanni .................. 56Leipzig .................... 35,90Lohengrin ..................... 49Mozart, W.A. .................. 11Mozart festival, see WürzburgWagner, R. .................... 22Vienna ..................... 11,42Würzburg ...................... 77
Don Giovanni .................. 56Leipzig .................... 35,90Lohengrin ..................... 49Mozart, W.A. .................. 11Mozart festival, see WürzburgWagner, R. .................... 22Vienna ..................... 11,42Würzburg ...................... 77
Topics
Example: Back-of-the-Book Index
Don Giovanni .................. 56Leipzig .................... 35,90Lohengrin ..................... 49Mozart, W.A. .................. 11Mozart festival, see WürzburgWagner, R. .................... 22Vienna ..................... 11,42Würzburg ...................... 77
Occurrences
Example: Back-of-the-Book Index
Don Giovanni .................. 56Leipzig .................... 35,90Lohengrin ..................... 49Mozart, W.A. .................. 11Mozart festival, see WürzburgWagner, R. .................... 22Vienna ..................... 11,42Würzburg ...................... 77
Differenttopic classes
Example: Back-of-the-Book Index
Don Giovanni .................. 56Leipzig .................... 35,90Lohengrin ..................... 49Mozart, W.A. .................. 11Mozart festival, see WürzburgWagner, R. .................... 22Vienna ..................... 11,42Würzburg ...................... 77
Differentoccurrences classes
Example: Back-of-the-Book Index
Don Giovanni .................. 56Leipzig .................... 35,90Lohengrin ..................... 49Mozart, W.A. .................. 11Mozart festival, see WürzburgWagner, R. .................... 22Vienna ..................... 11,42Würzburg ...................... 77
Example: Back-of-the-Book Index
Multiple Topic Names
Don Giovanni .................. 56Leipzig .................... 35,90Lohengrin ..................... 49Mozart, W.A. .................. 11Mozart festival, see also WürzburgWagner, R. .................... 22Vienna ..................... 11,42Würzburg ...................... 77
Association
Example: Back-of-the-Book Index
Book ContentTV Content Music content
MR. M. Random HouseBOL.com
Maria
C. SantanaThe Firm
Grisham DVDs
InterviewAuthorship
Publishing
Composership
Selling
Topic Map Concepts
Resources
Topics
Occurrencelinks
Associations
Assoc. classes
Subject Subject
Subject
SubjectSubject Subject
Subject
Subject
Book ContentTV Content Music content
More Concepts
Occurrenceclasses
Topic classes
MR. M. Grisham
The Firm
Random HouseSanatana
Maria
BOL.com DVDs
Person
Thriller
Online shopPublisher
BandSong
Media
Article
Biography
Latest bestseller
Latest hit
Book ContentTV Content Music content
More Concepts
Super classes
MR. M. Grisham
The Firm
Random HouseSanatana
Maria
BOL.com DVDs
Person
Thriller
Novel
Literature
AuthorJournalist
Association Concepts
Tina
Jim
Father John
Wife
Husband
Priest
Marriage
Role playingtopics
Associationinstance
Associationclass
Assoc. roles
Book ContentTV Content Music content
More Concepts
Scopes
MR. M. Grisham Random HouseSanatana
Maria
BOL.com DVDs
The Firm (English)Die Firma (Deutsch)
Book ContentTV Content Music content
More ConceptsIdentity
MR. M. Grisham
The Firm
Random HouseSanatana
Maria
BOL.com DVDs
http://www.topicmaps.org/PSI/authors.html#john-grisham
John Grisham
Summary: Topic Map Concepts
n Topic (reified subject)
n Occurrence
n Association, association role
n Topic class, occurrence class, and association class
n Class-instance
n Super-subclass
n Scope and scoping topic
n Identity and subject indicator
Scene III:
The Family of Topic Map Standards
ISO/IEC 13250, Data Model, Conceptual Model, TMQL, TMCL
Family of TM Standardsn ISO/IEC 13250:2000n ISO standard defining general concepts and interchange
syntax (SGML/HyTime + XML/Xlink)
n TM Data Modeln ISO project
n The foundation of the TM paradigm
n Independent of any particular (storage/interchange) syntax
n TM Conceptual Modeln ISO project
n Defines mapping between particular syntax (SGML and XML) and TM Data Model
Family of TM Standards cont‘d
n TMQL – TM Query Language
n ISO project
n ‘SQL’ for TMs
n Standardized creation/modification of TMs stored in TM Management Systems
n TMCL – TM Constraint Languagen ISO project
n Framework for the definition of ontologies / schemas for vertical or domain specific applications
n Support for semantic validation
Scene IV:
Topic Maps and Related Paradigms
Semantic Networks, RDF
Topic Maps and Semantic Networksn Pros of Semantic Networks:n Inheritance of node properties
n Inferencing
n Partitioning
n Formal notation
n Pros of TMs:n Occurrences
n Rich associations (n-ary, roles)
n Subject Identity
n Merging
n Standardized notation
Topic Maps and RDF
n TM / RDF – Similarities
nStructured, complex metadata
nBased on graphs
nStandardized notations
nKnowledge representation, ontologies
nHelp power the Semantic Web idea
n TMs on top of RDF ó RDF on top of TMs
Topic Maps and RDF cont‘d
n TM / RDF – Differences
‘toys’ (as of today)real products, projects, use
–merging
–distinguishes between addressable and non-addressable subjects
directed binary relationsn-ary associations with role players (instead of direction)
simple data structurepre-defined semantics
resource-centrictopic-centric
RDFTM
Act II:Allegro
TMs and knowledge representation
Quine‘s Criterion
What is there?
Quine‘s Criterion
Everything!
Scene V:
Ontologies, Schemas, Templates
The Starting Point …
TM Ontology
n John F. Sowa:
“Ontology defines the kinds of things that exist in the application domain.”
or
“A classification of the types and subtypes of concepts and relations necessary to describe everything in the application domain.”
“Real” TM
Infopool
is in is in takes place in
Austin
KnowledgeTech. 2001
U.S.A. Texas
Ontology TM
Topic classes Occurrence classes Association classes
country
state
city
conference
is in
takes place in
article
call f. papers
city map
video
Others:Assoc. roles,scoping topics
Ontology TM
Topic classes Occurrence classes Association classes
country
state
city
conference
is in
takes place in
article
call f. papers
city map
video
Others:Assoc. roles,scoping topics
These are all
These are all
Topics!
Topics!
Solution
n Terms TM Ontology, TM Constraint, TM Template, and TM Schema were coined by ISO committee
n Cover all TM constructs which have a declarative meaning:nAll classes and scoping topics
nClassification (see later)
nConstraints (see later)
=> ISO initiative TM Constraint Language (TMCL)
Scene VI:
Class Hierarchies
Building blocks are part of XTM
HUMAN
is instance of
Requirements
SPECIES
is instance of
MAMMAL
is subtype of
Graham Topic instance
is instance of
Requirements cont‘d
MAMMAL
Graham Topic instance
Requirements cont‘d
SPECIES
Graham Topic instance
is instance of
MAMMAL
is NOT aninstance of
Examples
n Topic classes:
Object →
piece of art →
painting, sculpture, novel, poem, opera
Person →
artist →
painter, sculptor, writer, poet, composer
Examples cont’d
n Association classes:
Object “fostered by” person →
piece of art “created by” artist →
opera “composed by” composer
Scene VII:
Inferencing
Deducing knowledge …
Association Properties
n Assoc: geo_object is in geo_object
n Properties: transitive, anti-reflexive, and anti-symmetric
n Facts in TM:Bavaria is in GermanyWürzburg is in BavariaMunich is in Bavaria
n Derived knowledge:Würzburg is in GermanyGermany is not in Bavaria
Inference Rules
n Class hierarchies and transitivity allow deduction of knowledge not explicitly coded in TM
n But TM might contains more knowledge which could be derived
n Inference rules define – as part of the ontology – how to derive further knowledge
Example
If $topic1 is sibling of $topic2 and $topic1 is male
then $topic1 is a brother
(Eric Freese, XML Europe 2000, Paris)
Rule components
n “if <condition> then <inference>” defines the inference rule
n “$topic1” and “$topic2” are variables which have to be instantiated when the rule is evaluated
n “is a sibling of” and “is a male” are the assoc. types in question
n “is a brother” is the inferred assoc. type
Scene VIII:
Consistency constraints
Dealing with millions of topics …
The Needs
nManual checking of large TMs is impossible but validation is a requirement
n TM software should validate during design and creationnPermanently or on demand
n Like structure validation in SGML/XML editors/parsers
n Constraints control validation process
=> ISO initiative TM Constraint Language (TMCL)
Examplen Topic type constraints:n Names (scope, number)
n Occurrence role (scope, number)
n Plays certain role in an association
n Example:n Person
n min. 1 englisch basename
n biography (exactly 1), portrait (max. 1)
n participates in born-in association
Example cont’d
n Association type constraints:n Scope
n Association role (number)
n Topic types of associated topics
n Example:n is in
n 1 Containee 1 Container
n city country, state, countycounty country, statestate country
Scene IX:
Topic Map Query Language (TMQL)
Query and modify TMs in a Standardized Manner
TMQL Sound Bites ...
n “Make Topic Maps Operational”
n “SQL for Topic Maps”
n “Backbone of Global Knowledge Interchange”
TMQL Applied to a 3-Tier Architecture
RDBMS TM MS
Business Logic
SQL TMQL
User Interface
TMQL
TMQL
XTM/13250
TMQL
TMQL
XTM/13250
TMQL System Context
XTM/13250
Application
Application
TMQL
TMQL
To
pic
Map
MS
Topic Mapdata model
Interchange syntax interface
TMQLinterface
Epilogue:
Conclusions
Topic Maps and the Ontological World
Conclusions
n TMs provide a simple but powerful paradigm
n Real products and real projects and real productive use 18 months after publication of ISO standard
n Accompanying standards (TMCL, TMQL) makeTMs ready for the Semantic Web and KM applications
n Harmonization with RDF
n But: TMs don’t aim at “Heavy Ontologies” (yet)
Some Resources
Addressable and Non-Addressable
Resourcesn Addressablen http://www.topicmaps.org
n http://www.infoloom.com/mailman/listinfo/topicmapmail
n http://k42.empolis.co.uk
n Non-addressablen Standardization:
ISO JTC1 SC34 WG3n Vertical applications:
OASIS Member Section TopicMaps.Org and its various Technical Committees
Announcement:empolis k42® EGP
Free access for research projects to k42
Announcement
empolis k42® Education Grant Programgives non-profit research projects free access to its k42 Knowledge Server
Find more athttp://k42.empolis.co.uk/egp.html
</End>
Thank You!
Any Questions?
empolisBERTELSMANN MOHN MEDIA GROUP
Transforming Information into Value