Exploiting Links to Improve Search in XML Documents

Abstract

This paper describes an approach that exploits links in XML retrieval. The proposed approach consists of reranking the set of documents returned for a given query by considering three sources of evidence namely, the relevance scores w.r.t query of a document neighbours, the text of the anchor links, and the document title tag. Our approach is evaluated on INEX 2006 collection. The results showed significant improvements of the retrieval performances. Subject Categories and Descriptors I.2.7 Natural Language Processing] Language Generation [I.7.2 Document Preparation] Markup languages General Terms: Document Processing; XML; Query Processing