Using Redescriptions and Formal Concept Analysis for Mining Definitions Linked Data

Justine Reynaud (1) , Yannick Toussaint (1) , Amedeo Napoli (1)

1 ORPAILLEUR - Knowledge representation, reasonning

In this article, we compare the use of Redescription Mining (RM) and Association Rule Mining (ARM) for discovering class definitions in Linked Open Data (LOD). RM is aimed at mining alternate descriptions from two datasets related to the same set of individuals. We reuse RM for providing category definitions in DBpedia in terms of necessary and sufficient conditions (NSC). Implications and AR can be jointly used for mining category definitions still in terms of NSC. In this paper, we firstly, recall the basics of redescription mining and make precise the principles of definition discovery. Then we detail a series of experiments carried out on datasets extracted from DBpedia. We analyze the different outputs related to RM and ARM applications, and we discuss the strengths and limitations of both approaches. Finally, we point out possible improvements of the approaches.

Mots clés: Concept Analysis; Association Rule Mining; Linked Open Data; Definition of categories; Redescription Mining

Domaines: Informatique [cs]

Justine Reynaud, Yannick Toussaint, Amedeo Napoli. Using Redescriptions and Formal Concept Analysis for Mining Definitions Linked Data. ICFCA 2019 - 15th International Conference on Formal Concept Analysis, Jun 2019, Francfort, Germany. pp.241-256. ⟨hal-02170760⟩