We present in this paper the work of the Information Retrieval Modeling Group (MRIM) of the Computer Science Laboratory of Grenoble (LIG) at the INEX 2007 Ad Hoc Track. We study here the impact of non structural relations between structured document elements (doxels) on structured documents retrieval. We use existing links between doxels of the collection, encoded with the collectionlink tag, to integrate link and content aspects. We characterize the relation induced by the collectionlink tags with relative exhaustivity and specificity scores. As a consequence, the matching process is based on doxels content and these features. Results of experiments on the test collection are presented. Runs using non structural links overperform a baseline without such links.
展开▼
