Similarity measure to identify users' profiles in web usage mining - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2010

Similarity measure to identify users' profiles in web usage mining

Résumé

Nowadays, content available on the Internet is continuously growing. Websites aregathering more and more information. It makes the website browsing process even harder. This paper addresses the web usage mining problem. We try to use a characteristic patterns algorithm to identify users' profiles. Our experiment shows that this algorithm is not suitable for our encyclopedic hypermedia. Therefore, we present an alternative approach through casting a user trace (set of transitions) as a graph. Then we suggest a similarity measure among different browsing traces. To validate this measure we used the t-SNE algorithm (t-Distributed Stochastic Neighbor Embedding) which allows us to project our data in a two dimensional space. Then we apply SVM classification algorithm (Support Vector Machine) and compare our results with the results of characteristic patterns algorithm.
Fichier principal
Vignette du fichier
2010_ABOU_LATIF.pdf (338.53 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-00560096 , version 1 (27-01-2011)

Identifiants

  • HAL Id : hal-00560096 , version 1

Citer

Firas Abou Latif, Nicolas Delestre, Nicolas Malandain, Jean-Pierre Pécuchet. Similarity measure to identify users' profiles in web usage mining. INFORSID XXVIII°, May 2010, Marseille, France. pp.77-92. ⟨hal-00560096⟩
144 Consultations
155 Téléchargements

Partager

Gmail Facebook X LinkedIn More