Knowledge Discovery and Enrichment from Scholarly Data for Expert Finding

International audienceWith the generalisation of the digitalization of scientific publications , scholarly data is now tackled with a big data perspective. In this context, interest about applications like expert finding, research recom-mandation systems or collaborators discovery grows and challenges related to knowledge discovery on large-scale scholarly data arise. The Unified Knowledge Platform (Plateforme de Connaissances Unifiées) project aims at developing an open-source platform valuing scholarly as well as business data. In that respect, we propose an approach and a methodology for discovering knowledge and enrich it from workings documents, more precisely scientific publications, in the particular use case of expert finding. Compared to the state of the art, the originality of our approach lies in the combination of text mining as well as graph mining methods, more specifically graph abstraction. We present an experiment with already published results on the 9-years acts of a French workshop on semantic information retrieval. In this experiment, we managed to obtain a graph mapping researchers who participated in the workshop. In this graph, researchers are linked together by co-publication relationships and described by their topics of publication. We were also able to detect dense communities of researchers with the help of graph abstraction. Based on these results and in the light of the state of the art, we discuss further research tracks