In information retrieval on classical structured documents, one problem consists in browsing the result space using the structure of the documents. Taking into account other links between doxels increases this problem. In this article, we consider relative exhaustivity and relative specificity values computed on non compositional linked doxels to index the corpus ; adding this information to th...

Résumé : Dans le cadre de recherches menées sur l’indexation sémantique, nous avons été conduits à nous interroger sur l’efficacité générale des modèles actuels de représentation des terminologies au sein des ontologies. Après avoir évoqué leurs limites actuelles, nous proposons un nouveau modèle pour manipuler une ressource termino-ontologique en OWL et nous en décrivons son implémentation pro...

Our work is in the field of the support systems to museum visits and access to cultural heritage. Our goal is to design a recommendation system implemented on mobile devices adaptable to the user’s profile and sensitive to its context, to improve the user experience and help build his visits to the museum based on his preferences and constraints. MOTS-CLÉS : systèmes de recommandation, web séma...

We describe in this article an annotation scheme for deep dependency syntax, built from the surface annotation scheme of the Sequoia corpus, abstracting away from it and expressing the grammatical relations between content words. When these grammatical relations take part into verbal diatheses, we consider the diatheses as resulting from redistributions from the canonical diathesis, which we re...

Sentiment classification and recommender systems were until recently completely disjoint domains. Recommender systems exploit the users/items/rates matrix with omitting the available text information. Sentiment classification exploits text reviews and consumers rates to build models for document analysis. In this article we propose an unified model exploiting both text and user, items and rates...

We present a comparative study on the impact of the nature and size of the training corpus on performance in automatic named entities recognition. This evaluation is in the form of multiple modulations on three French corpus. Two corpora are from the catalog of the European Language Resources Association (ELRA) and the third is composed of documents extract from the OpenEdition.org platform. Mo...

This paper presents a generic ,modular and interactive architecture for embodied conversational agent called Greta. It is 3D agent able to communicate with users using verbal and non verbal channels like gaze, head and torso movements, facial expressions and gestures. Our architecture follows the SAIBA framework that defines modular structure, functionalities and communication protocols for ECA...

what is the impact of the transcription on the phonetization This paper aims at quantifying the impact of the transcription enrichments on the automatic phonetization of speech. Experiments were carried out on a 7 minutes French corpus including conversational speech, readed speech and a political discourse. Results showed the better the transcription the better the phonetization and that indep...

This paper studies the problem of images indexing and retrieval related to museum visits. We especially focus on "offline" use of mobile devices (i.e., without connection to a remote server), from the point view of intrinsic quality and the point of view of mobile potential use. We describe three approaches, and we study their qualitative behavior on a test collection of photographs of painting...

OLAP and Data Warehouse (DW) systems are decision support technologies allowing users to explore and analyze huge volume of data. Benefits of conceptual modeling of DW have been widely recognized in several works. In this position paper we address stream data (i.e. real real-time continuous ordered and sequence of items), and we define the requirements of conceptual models for DW integrating st...