Indexation de blocs extraits de pages Web en utilisant le rendu visuel

This paper presents a Web page indexation model. In this model, a Web page is not viewed as a whole, but as a combination of a set of blocks based on their visual rendering, where each bloc shares is own semantic. The indexation of a page Web is achieved in two steps : (1) construction of a hierarchical tree of visual blocks based on block visual layout in the Web page (2) textual indexation of each block by a term vector and taking into account blocks importance and indexation of neighbouring blocks (parent, children, siblings...). MOTS-CL

• Type: Conference
• Year: 2008
• Where: CORIA
• Authors: Nicolas Faessel
• Keywords: CORIA 2008, Indexation, Information Technology, Web Page, Web Page Indexation