No abstract

Abstract. In a context of decision-aid to support the identification of collaborative networks, this paper focuses on extracting essential facets of firm competencies. We present an approach for enrichment of competence ontology, based on two steps where a novel effective filtering step is utilized. First we extract the correlation between terms of a learning dataset using the generation of association rules. Second we retain the relevant new concepts using an extracted semantic information. The suggested approach was tested on an ontology of mechanical industry competencies. Experiments were performed on real data, which show the usefulness of our approach.

L'extraction automatique d'opinions sur le web 2.0 est un domaine de recherche de plus en plus étudié. Elle utilise souvent deux méthodes à vocations différentes : soit des méthodes fondées sur l'apprentissage par la constitution de corpus en vue d'établir des modèles pour la classification, soit rechercher des mots caractéristiques tels que les adjectifs qui contribueront à la classification des textes. Dans ce dernier cas, les outils existants utilisent des dictionnaires généraux, et possèdent des limites : pour certains domaines, des adjectifs peuvent être inexistants voire contradictoires. Dans cet article, nous proposons une nouvelle approche de création automatique de dictionnaire d'adjectifs intégrant la connaissance du domaine. Les expériences menées sur des données réelles ont montré l'intérêt de notre approche comparativement à une méthode plus classique par apprentissage. ABSTRACT. Expressed opinions grows more and more on the Internet. Recently, extracting automatically such opinions becomes a topic addressed by new research work. Traditionally, detection of opinions is based on extracting adjectives. Existing methods are often based on general dictionaries. Unfortunately, main drawbacks of these approaches are that, for different domains, adjectives could not exist and could have an opposite meaning. In this paper we propose a new approach to the automatic creation of dictionary of adjectives that integrates the domain knowledge. The experiments conducted on real data show the usefulness of our approach, compared to a more classic method based on machine learning mechanisms.

No abstract

Most question and answer systems"Q&A" are based on three research themes: question classification and analysis, document retrieval and answer extraction. The performance in every stage affects the final result. The classification of questions appears as an important task because it deduces the type of expected answers. A method of improving the performance of question classification is presented, based on linguistic analysis (semantic, syntactic and morphological) as well as statistical approaches guided by a layered semantic hierarchy of fine grained question types. Actually, methods of question expansion are studied. This method adds for each word a higher representation. Various features of questions, diverse term weightings and several machine learning algorithms are compared. Experiments were conducted on real data are presented. They demonstrate an improvement in precision for question classification.