Section: New Results

Ontology matching and alignments

We pursue our work on ontology matching and alignment support with contributions to evaluation and alignment semantics.

Evaluation

Evaluation of ontology matching algorithms requires to confront them with test ontologies and to compare the results. Since 2004, we run the Ontology Alignment Evaluation Initiative (oaei ) which organises evaluation campaigns for assessing the degree of achievement of actual ontology matching algorithms [2] .

This year, we ran two evaluation campaigns named 2011.5 and 2012. This was justified by the will to complete full evaluations using the support of the seals platform. Hence, the main activities carried out in 2012 were related to the automation of the evaluation. This involved providing participants with a better way to bundle their tools so that they can be evaluated both offline and within the seals platform. It also required to support more organisers to provide test case within the plaform.

This work has been used in the oaei 2012 evaluation campaign. oaei 2012 offered 9 different test sets (7 of which under the seals platform). This issue brought the following results:

• More participants than ever (21);

• All ontology matchers running on the SEALS platform (18);

• Increased performances in terms of precision and recall;

• Matchers are now very scalable and can deal with the largest available ontologies (9 systems able to deal with the very large medical ontology SnoMed);

We have also introduced as a data set, the benchmark for multilingual ontology matching developed last year [6] . It has pushed matcher developers to address multilingual issues.

The participating systems and evaluation results were presented in the 7th Ontology Matching workshop, that was held in Boston, MA, US [22] , [7] . More information on oaei can be found at http://oaei.ontologymatching.org/ .

Semantics for weighted correspondences

Alignment correspondences are often assigned a weight or confidence factor by matchers. Nonetheless, few semantic accounts have been given so far for such weights. We have proposed a formal semantics for weighted correspondences between different ontologies. It is based on a classificational interpretation of correspondences: if $o$ and $o^{\text{'}}$ are two ontologies used to classify a common set $X$, then alignments between $o$ and $o^{\text{'}}$ are interpreted as encoding how elements of $X$ classified in the concepts of $o$ are re-classified in the concepts of $o^{\text{'}}$, and weights are interpreted as measures of how precise and complete re-classifications are. This semantics is justifiable for extensional matchers. We have proven that it is a conservative extension of a semantics of absolute correspondences, and we have provided properties that relate correspondence entailment with description logic constructors [8] .

This work has been made in cooperation with Alexander Borgida (Rutgers University) and Chiara Ghidini and Luciano Serafini (Fondazione Bruno Kessler).