Overall Objectives
Scientific Foundations
Application Domains
New Results
Other Grants and Activities

Section: New Results

Ontology networks

Dealing with the semantic web, we are interested in ontology networks, i.e., sets of distributed ontologies that have to work together. One way for these systems to interact consists of exchanging queries and answers. For that reason, we pay particular attention to query systems.

Path queries and μ-calculus

Participants : Melisachew Wudage Chekol [Contact] , Jérôme Euzenat, Pierre Genevès, Nabil Layaïda.

In the continuation of our previous work on path-based RDF querying and WAM 's work on μ-calculus interpretation of XPATH , we are using the same techniques for interpreting SPARQL queries over RDF .

To that extent, we proposed a μ-calculus encoding of RDF graphs and SPARQL queries. They allows to translate query evaluation to graph traversing through the modalities of the logics. μ-calculus offers the opportunity to encode more interesting contexts such as querying with path or querying module ontologies. We are studying the query containment and equivalence problem for various forms of SPARQL fragments, both from the standpoint of complexity bounds and effective evaluation strategies.

Trust in peer-to-peer semantic systems

Participants : Manuel Atencia [Contact] , Jérôme Euzenat, Marie-Christine Rousset.

In a semantic peer-to-peer network where peers resort to different ontologies, links between peers are then realised by means of alignments. So queries (and their answers) are successively translated before peers receive them. However, even when alignments have been computed with human intervention, there is no guarantee that peers will obtain satisfactory answers to their queries.

A trust mechanism can assist peers to select those paths in the network that are better suited to their queries. We have put forward a probabilistic approach for trust which takes into account ontological information, alignments and past experiences to compute trust values. Bayesian inference is performed when approximating probability values.


Logo Inria