Modélisation multi-agents d'un système de recherche d'information multimédia à forte composante vidéo

Abstract : Techniques to retrieve and present multimedia documents require heterogeneous knowledge and know-how (librarians, media specialists, application domain specialists, etc). Most of these techniques are quickly evolving, as video processing is an example. We are particularly interested in video retrieval. Librarians searching for documents traditionally use some surrogates of them, represented by lists of keywords. We adopt the conceptual graph theory of Sowa to represent documents descriptions. In our model the descriptions are modelized by facets, each facet matching the viewpoint of an expert on the collection. During the retrieval process, documents descriptions can be exploited by a vectorial or a conceptual graph matching, depending on the user's need. Our approach consists in conceiving a set of agents to assist users in their information retrieval. This "society of agents" must be able to adjust its behaviour to various information needs. We define the main agents of such a system and their organisation, studying their interactions to improve the retrieval. This model has been implemented a multiagent prototype, Video-SAFARI.
Document type :
Theses
Complete list of metadatas

Cited literature [154 references]  Display  Hide  Download

https://hal.univ-lorraine.fr/tel-01748617
Contributor : Thèses Ul <>
Submitted on : Wednesday, June 20, 2018 - 2:04:44 PM
Last modification on : Tuesday, January 29, 2019 - 1:21:51 AM
Long-term archiving on : Tuesday, September 25, 2018 - 7:22:19 AM

File

SCD_T_1996_0017_SIMONNOT.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : tel-01748617, version 1

Collections

Citation

Brigitte Simonnot. Modélisation multi-agents d'un système de recherche d'information multimédia à forte composante vidéo. Autre [cs.OH]. Université Henri Poincaré - Nancy 1, 1996. Français. ⟨NNT : 1996NAN10017⟩. ⟨tel-01748617⟩

Share

Metrics

Record views

65

Files downloads

81