Skip to Main content Skip to Navigation

Une nouvelle approche topologique pour la recommandation de tags dans les folksonomies

Abstract : We focus in this thesis on the problem of tag recommendation in social sharing to classification systems called folksonomies. Users of a folksonomy annotate their resources with freely tags chosen. We propose here a new topological approach for tags recommendation called TLTR (Two Level Tag Recommendation). TLTR (Two Level Tag Recommendation) is based on an original approach of graph compression. The graph of a folksonomy is compressed by a clustering each of the three components, namely the set of users, resources and tags. A topological clustering method based on a seed-centered approach for community detection in multiplex graphs is proposed. A classical topological approach, namely Folkrank, is applied to the reduced graph to select the most appropriate clusters of tags. These clusters are then used to build another contextual graph extracted from the original graph representing the folksonomy. Folkrank method is applied again to compute the list of tags to recommend. Experiments on large folksonomy, including, data extracted from references system Bibsonomy show the relevance of our approach.
Complete list of metadatas

Cited literature [4 references]  Display  Hide  Download
Contributor : Abes Star :  Contact
Submitted on : Tuesday, March 20, 2018 - 6:11:08 PM
Last modification on : Tuesday, June 2, 2020 - 7:54:02 AM
Long-term archiving on: : Tuesday, September 11, 2018 - 11:27:12 PM


Version validated by the jury (STAR)


  • HAL Id : tel-01739216, version 1


Manel Hmimida. Une nouvelle approche topologique pour la recommandation de tags dans les folksonomies. Linguistique. Conservatoire national des arts et metiers - CNAM, 2015. Français. ⟨NNT : 2015CNAM1054⟩. ⟨tel-01739216⟩



Record views


Files downloads