From Thesauri to Ontologies Experiences from MuseumFinland Semantic Portal and National Ontology Pro

1
From Thesauri to Ontologies Experiences from
MuseumFinland -Semantic Portal and National
Ontology Project in Finland

• Mirva Salminen
• email mirva.salminen_at_helsinki.fi
• University of Helsinki, Dept. of Computer
  Science,
• and the Helsinki Institute for Information
  Technology (HIIT)
• Semantic Computing Research Group
• http//www.cs.helsinki.fi/group/seco/

05.05.2004
2
Contents of presentation

• National ontology project in Finland
• Ontologies
• Solving problems of thesauri
• Advantages of content description with ontologies
• Use of ontologies MuseumFinland application
• Conclusions

3
National Ontology Project in Finland

• Establishing the semantic infrastructure for the
  Finnish semantic web
• Motivations for the project
• Thesauri and classification systems have been
  widely used for indexing contents
• There are, however, many problems involved!
• Current thesaurus-based approach is not good
  enough for the semantic web
• Machine understandability missing
• Funded by Tekes and a large consortium of
  companies and institutions
• 9/2003-8/2005

4
Research consortium
The Finnish Museum of Photography
TSK
Connexor Oy
Finnish National Gallery
5
Contents of presentation

• National ontology project in Finland
• Ontologies
• Solving problems of thesauri
• Advantages of content description with ontologies
• Use of ontologies MuseumFinland application
• Conclusions

6
Ontologies

• An ontology is an explicit specification of a
  conceptualization (Gruber, 1995)
• A conceptualization is a set of conceptual
  relations defined on a domain space (Guarino,
  1998)
• Expressed in a formal machine understandable way
  (e.g. RDF(S), OWL)
• Examples WordNet, Standard Upper Ontology (SUO),
  dmoz.org, TAP, MAO, FRBR, CIDOC CRM,
• Define vocabulary for metadata formats
• e.g. Dublin Core, LOM,

7
Example 1 Ontology
Place
Action
Object
Actor
Textile factory
Weaving
Fabric
Weaving machine
Weaver
productOfAction
placeOfAction
toolOfAction
actorOfAction
8
Why use ontologies in content description?

• Ontologies overcome problems related to thesauri
• Use of ontologies has advantages for content
  description

9
Contents of presentation

• National ontology project in Finland
• Ontologies
• Solving problems of thesauri
• Advantages of content description with ontologies
• Use of ontologies MuseumFinland application
• Conclusions

10
Problems of thesauri

• Interoperability
• Identification of concepts
• Semantics too simple
• Managing large thesauri
• Managing changes

11
1. Interoperability

• Systems are hetegoneous
• Syntax systems use different data syntax/models
• Semantics Terminology of different application
  fields, organizations and catalogers differ
• The result Systems cannot operate together
• Ontologies are expressed in a formal machine
  understandable way (RDF(S), OWL)
• Formal languages can be automatically translated
  into other formal languages using existing tools

12
2. Identification of concepts

• In thesauri homonymy need to be distinguished
• Polysemy is hard to disambiguate
• Homonymic words can be distinguished with URI
  (Uniform Resource Identifier)
• Corresponding concepts can be disambiguated by
  their context
• Can be applied to meronymic terms as well

13
Example Handling homonymies
Building
Edge
River
Lloyds
Bank
Bank
RiverBank
BankBuilding
14
3. Semantics too simple

• The semantic system (NT, BT, RT, etc.) is too
  simple for creating truly intelligent systems
• Dealing with uncertainty and fuzzy concepts
• E.g. meronymies, different associative relations
  etc. are needed
• Ontologies provide a way to define new relations
  when needed.
• Relations can also be defined in a more detailed
  manner name, domain, range, symmetric/transitive
  etc.
• Ontologies make it possible to also have grouping
  concepts that do not mix with the real terms.

15
4. Managing large thesauri

• No organization is capable of maintaining the
  thesauri of all fields
• The work has to distributed to different expert
  groups working together
• Distributed maintenance of ontologies can be
  assisted by computers
• Ontologies can also be mapped with other
  ontologies reaching a large web of interoperable
  semantics where all the individual ontologies are
  maintained by the experts of that area
• Open source would boost application development

16
Example Managing large ontologies
Eucaryotes
Actor
Animals
Association
Person
Vertebrates
Institution
Man
Mammals
Fire brigade
Army
Man
17
5. Managing changes and time

• The thesauri and concepts change over time
• New concepts emerge all the time, Czechoslovakia
  does not exist any more, Petsamo is not part of
  Finland today, etc.
• The contents are indexed with old
  keywords/concepts but may be retrieved with new
  ones
• Ontology versions can be made interoperable by
  bridging them annotated objects can be found
  through both the old and the new concepts
• Validity time can be attached to the concepts and
  logics can be used to reason about them
• No confusion about changes in time, e.g. Petsamo
  of today is not optimal for annotating object
  dating to 1930s

18
Example Managing changes and time
Ontology version 1
Change bridge
Ontology version 2
East Germany
East Germany
Germany
Germany
West Germany
West Germany
1949 - 1990
1990
1990 -
2004
1949
1990
19
Contents of presentation

• National ontology project in Finland
• Ontologies
• Solving problems of thesauri
• Advantages of content description with ontologies
• Use of ontologies MuseumFinland application
• Conclusions

20
Ontologies in content description

• Advantages
• Better usability
• Web of concepts at reach
• Enrichment with semantic inference rules

21
1. Better usability and computer aided metadata
creation

• Better semantic content would make it possible to
  create more intelligent user interfaces
• Typing in keywords and reading the hit list is
  not the only possibility!
• Ontologies enable semantic browsing, view-based
  search, graphical interfaces, content
  visualization etc.
• Computer helps in choosing the right concept for
  content description fancier browsing of concepts
  and automatic limitation of suggested concepts

22
2. Web of concepts at reach

• Content need not be described as precicely as
  with words
• Once annotating the concept with some ontological
  concepts, the object is linked to the whole
  ontology and can be searched through all the
  links between concepts in the ontology

23
3. Enrichment with semantic inference rules

• Ontologies can be further enriched with logical
  inference rules.
• Logical rules define new and more complex
  relations between ontological concepts
• Objects annotated with the data inherit these
  rules and can be searched through these rules
  between concepts in the ontology

24
Example Web of concepts 1
Eucaryotes
Animals
Fungi
Vertebrates
Mammals
Dinosaurs
Modern birds
25
Example Web of concepts 2
26
Example Web of concepts 3
s
27
Contents of presentation

• National ontology project in Finland
• Ontologies
• Solving problems of thesauri
• Advantages of content description with ontologies
• Use of ontologies MuseumFinland application
• Conclusions

28
MuseumFinland

• Case study and demonstration of the possibilities
  of the Semantic Web
• A semantic portal for Finnish museums to publish
  their collections together on the Semantic Web
• A result of the Semantic Web research project
  on publishing content on the web
• Duration 3/2002-3/2004
• Public pilot version from March 8,
  2004http//museosuomi.cs.helsinki.fi/

29
The Vision of MuseumFinland

• Global View to Distributed Collections
• One seamless national collection (virtually)
• Museums in Finland -gt Museum of Finland
• Intelligent Services to End-Users
• Search Concept-Based Information Retrieval
• Browsing Semantically Linked Contents
• Easy Content Publication for Museums
• Creating a national platform and a process for
  the museums to publish their content together on
  the Semantic Web

30
Demonstration of MuseumFinland
http//museosuomi.cs.helsinki.fi/
31
Contents of presentation

• National ontology project in Finland
• Ontologies
• Solving problems of thesauri
• Advantages of content description with ontologies
• Use of ontologies MuseumFinland application
• Conclusions

32
Conclusions

• Ontologies provide a basis for content
  descriptions that is more flexible than thesauri
• Formal and exact semantics of ontologies enable
  the creation of intelligent applications
• Ontologies are supported by new WWW standards
  (Semantic Web)
• Content publication, interoperability

33
Sources

• Gruber, T. R. Toward Principles,for the Design of
  Ontologies Usedfor Knowledge Sharing. Int. J.
  Human-Computer Studies, 43 (1995) 907-928.
• Guarino, N. Formal Ontology and Information
  Systems. Formal Ontology in Information Systems.
  10s Press, 1998.
• E. Hyvönen, M. Junnila, S. Kettula, E. Mäkelä, S.
  Saarela, M. Salminen, A. Syreeni, A. Valo, K.
  Viljanen MuseumFinland - Finnish Museums on the
  Semantic Web. User's Perspective. Proceedings of
  Museums and the Web 2004 (MW2004), March 31 -
  April 3, 2004, Arlington, Virginia, USA.

34
Thank you!
Mirva Salminen email mirva.salminen_at_helsinki.fi
University of Helsinki, Dept. of Computer
Science, and the Helsinki Institute for
Information Technology (HIIT) Semantic Computing
Research Group http//www.cs.helsinki.fi/group/sec
o/