Une Indexation conceptuelle pour un filtrage par dimensions

In order to resolve multi-dimentional queries, we propose a conceptual indexing based on a medical meta thesaurus (UMLS). We study the impact of this indexing compared to a words-based indexing. We show that using a meta thesaurus is delicate to set up, but can give better results than words-based indexing. Then, we define the notion of query dimensions. Exploiting the hierarchical organization of concepts of the meta thesaurus, we propose a simple technique to filter the corpus according to the query dimensions. We have performed evaluation on the ImageCLEFmed 2005 collection. Using conceptual indexing and taking into account the query dimensions, we improve results about 16%. MOTS-CL

• Type: Conference
• Year: 2006
• Where: CORIA
• Authors: Saïd Radhouani, Loïc Maisonnasse, Joo-Hwee Lim, Thi-Hoang-Diem Le, Jean-Pierre Chevallet
• Keywords: CORIA 2006, CORIA 2007, Meta Thesaurus, Query Dimensions, Words-based Indexing