Combinaison de critères par contraintes pour la Recherche d'Information Géographique

Abstract : Recent studies show an increasing proportion of queries with geographic criteria on Web search engines. This part is even bigger on specific corpora like cultural heritage collection (e.g. travelogues). We admit that the geographic information is composed of three facets: spatial, temporal and thematic. Works realized in our laboratory aim geographic information extraction from textual documents and the construction of independent and specific indexes for theses three facets. The goal of this thesis is to combine these three facets to support multicriteria searches. This work concerns several fields: Natural Language Processing (NLP), Geographic Information System (GIS), classic Information Retrieval (IR) and Geographic Information Retrieval (GIR). Our first contribution is about an original combination approach of specific indexes. During the retrieval process, it consists first in querying the different indexes independently and then combining the results lists. We propose also a user to personalize this combination with constraints. In order to realize this combination, we propose to imitate the homogenization approaches used in classical IR strategies that represent terms with corresponding lemmas. For geographic information, it consists in segmenting them on tiles and on using their occurrence frequency. So, our second contribution concerns a generic standardization approach implemented on spatial and temporal information. In order to evaluate these different propositions, we have tested and validated them via several prototypes and experimentations. The last contribution relates to an evaluation framework for GIR systems. Thanks to this framework, we verified and quantified the benefit of combining the different geographic information facets and also have compared several combination approaches.
Document type :
Theses
Complete list of metadatas

Cited literature [1 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00551889
Contributor : Damien Palacio <>
Submitted on : Monday, February 14, 2011 - 10:44:35 AM
Last modification on : Sunday, April 7, 2019 - 3:00:39 PM

Identifiers

  • HAL Id : tel-00551889, version 2

Collections

Citation

Damien Palacio. Combinaison de critères par contraintes pour la Recherche d'Information Géographique. Interface homme-machine [cs.HC]. Université de Pau et des Pays de l'Adour, 2010. Français. ⟨tel-00551889v2⟩

Share

Metrics

Record views

327

Files downloads

715