Regrouper des résultats SPARQL par comparaison de leurs contenus tels qu'ils sont agencés dans la base RDF interrogée

This paper proposes a novel approach to group SPARQL query results according to their similarities. In order to compare results, the originality of our approach is to consider for each result the data constituting this result such as they are present and arranged in the queryed RDF dataset. We therefore do not limit ourselves to comparing the results between them, but we recontextualize them in the dataset where they were selected in order to take into account not only the resources of results but also their links with others resources in the dataset, i.e. their neighborhoods. From the results and their neighborhoods, we define common descriptive patterns to hierarchically group results in order of similarity, thus providing the user with easier navigation and understanding of the results. MOTS-CLÉS : SPARQL, clustering query results, motif de description, hierarchical clustering.