Clémentine Adam

Learn More
RÉSUMÉ. L'application de méthodes d'analyse distributionnelle pour calculer des liens de proxi-mité sémantique entre les mots est devenue courante en TAL. Toutefois, il reste encore beaucoup à faire pour mieux comprendre la nature de la proximité sémantique qui est calculée par ces mé-thodes. Cet article est consacré à la question de l'évaluation d'une(More)
In this paper, we aim at automatically identifying Elaboration. This relation is particularly difficult to spot since it does not have prototypical markers. Our approach focuses on an ambiguous syntactic pattern, the gerund clause, combined with lexical cues. This approach allows us to detect few but accurate cases of inner sentence Elaborations in our(More)
Une évaluation de l'impact des types de textes sur la tâche de segmentation thématique Résumé. Cette étude a pour but de contribuer à la définition des objectifs de la segmentation théma-tique (ST), en incitant à prendre en considération le paramètre du type de textes dans cette tâche. Notre hypothèse est que, si la ST est certes pertinente pour traiter(More)
RÉSUMÉ Dans cet article, nous confrontons une base distributionnelle construite à partir d'un corpus d'articles de revues de sciences humaines à des substitutions observées dans les journaux de requêtes du moteur interrogeant ce même corpus ; le recouvrement entre les deux types de données est important (59%). Ces résultats contribuent à deux pistes de(More)
Using distributional analysis methods to compute semantic proximity links between words has become commonplace in NLP. The resulting relations are often noisy or difficult to interpret in general. This paper focuses on the issues of evaluating a distributional resource and filtering the relations it contains, but instead of considering it in abstracto, we(More)
In this paper, we present an analysis based on linguistic and typographic features that allows for the identification of titles in web documents. We focus in particular on procedural texts. Identifying titles is a difficult task because ways of encoding them are very diverse. A number of titles are also incomplete because of context, we propose therefore a(More)
This paper brings a contribution to the field of discourse annotation of corpora. Using ANN-ODIS, a french corpus annotated with discourse relations by naive and expert annota-tors, we focus on two of them, Elaboration and Entity-Elaboration. These two very frequent relations are (a) often confused by naive annotators (b) difficult to detect automatically(More)
In this paper, we present an analysis based on linguistic and typographic features that allows for the identification of titles in web documents. We focus in particular on procedural texts. Identifying texts is a difficult task because ways pf encoding them are very diverse. A number of titles are also incomplete because fo context, we propose also a way to(More)
  • 1