Indexation libre et contrôlée d'articles scientifiques. Présentation et résultats du défi fouille de textes DEFT2012 (Controlled and free indexing of scientific papers. Presentation and results of the DEFT2012 text-mining challenge) [in French]

Controlled and free indexing of scientific papers Presentation and results of the DEFT2012 text-mining challenge In this paper, we present the 2012 edition of the DEFT text-mining challenge. This edition addresses the automatic, keyword-based indexing of scientific papers through two tracks. The first gives to the participants the terminology of keywords used to index the documents, while the second does not provide this terminology. The corpus is composed of scientific papers published in humanities journals, indexed by their authors. This indexing is used as a reference for the evaluation. The results have been evaluated in terms of micro-measures on the recall, precision and F-measure computed after keyword lemmatization. In the track giving the terminology of used keywords, the mean F-measure is 0.3575, the median is 0.3321 and the standard deviation is 0.2985 ; in the second track, the mean F-measure is 0.2055, the median is 0.1901 and the standard deviation is 0.1516. MOTS-CLÉS : Campagne d’évaluation, fouille de textes, indexation libre, indexation contrôlée, mots-clés, thésaurus.