Home

Themes
  Theme 1
  Theme 2
  Theme 3

Research Units
1) University of Modena and Reggio Emilia
2) University of Trento
3) University of Bologna
4) University of Roma Tre

Meetings

Deliverables

Publications

THEME 2

EMERGING SEMANTICS: SEMANTIC MAPPINGS DISCOVERY AMONG ONTOLOGIES

Units involved in this Theme

University of Modena and Reggio Emilia

University of Trento

University of Bologna

University of Roma Tre


Goals of this Theme

The first goal is the definition of a language for representing complex mappings among domain ontologies. Such a language will be used to represent all the elements of mappings which need in order to solve a query with respect to different ontologies, that is the query rewriting, and the identification of useful sources. Another goal is the analysis and the development of semantic mapping techniques between ontologies, including an assessment of the contribution that each technique may provide to the discovery of the different kind of mappings defined in language. Innovative techniques to discover mappings among domain ontologies will be defined. In particular, we will consider discovery techniques based both on the language semantics plus lexical chains and logic inference. Moreover, since the synthetic description of the contents (instance) is a relevant aspect of the project, we will analyze techniques to infer mappings exploiting the similarity among the information sources data. The last goal is the definition of a general architecture for the discovery and the management of mappings between ontologies, which includes the main conceptual modules used in the semantic mappings framework, such as Lexical resources (e.g. WordNet) and automatic reasoning modules.

 

 
Working Phases

Phase 1 (6 months: December 1, 2004 - May 31, 2005)

The research activity will focus on the analysis of the semantic mapping techniques at the state of the art and on their comparison with each other (product D2.R1). This activity will include an evaluation of each technique contribution to the mappings computation, within the specific contest of WISDOM where the mappings among the domain ontologies, integrating an information sources collection into a global virtual schema, are relevant. The matching algorithms proposed in literature will be analyzed to the ones including techniques to solve conflicts among the different representations of the same concept. Furthermore, standard proposals of ontologies mappings will be analyzed in order to evaluate their use within the WISDOM project. On the basis of the results of these analyses, we will develop a common framework to support domain ontology mapping, with specific reference to the definition of which kind of information has to be included in a mapping.

Phase 2 (6 months: June 1, 2004 - November 31, 2005)

The goal of this phase is twofold. First, we will define a language able to represent complex mappings among heterogeneous domain ontologies. Then, innovative techniques to discover mappings among domain ontologies will be defined. In particular, we will consider discovery techniques based both on the language semantics plus lexical chains (sources discovery) and logic inference (Context Matching). Moreover, since the synthetic description of the contents (instance) is a relevant aspect of the project (according with THEME1), we will analyze techniques to infer mappings exploiting the similarity among the information sources data.

Phase 3 (12 months: December 1, 2005 - November 31, 2006)

The first activity will address the extraction of the sources synthetic representation by means of the lexical chains technique. In particular, the algorithms, proposed in THEME 1, will be evaluated on the basis of different parameters: some technological (soundness of the extraction process, computational complexity), some others qualitative (lexical chains expressiveness to describe the sources, extension capability in the multi-linguistic area, techniques effectiveness with respect to the semantics assigned to the extracted data). Semantic similarity measures among lexical chains and the reference ontology will be studied and developed.
A software prototype implementing a platform to manage and discover semantic mappings among domain ontologies will be developed in collaboration with the units of Modena e Trento (product D2.P1). The prototype can be viewed as a service to be invoked to generate such mappings. It will constitute a highly modular system, domain independent, where the different components can be inserted “plug and play” or customized. On the basis of the architectural choices made in the WISDOM project, the prototype will be used as a global shared service, or locally used in “peer-to-peer” modality. The prototype will be tested by means of web information sources.


 

 

Mailing List :  wisdom@dbgroup.unimo.it

Webmaster: Enrico Ronchetti