GPX - Gardens Point XML IR at INEX 2006

Abstract

The INEX 2006 evaluation was based on the Wikipedia collection in XML format. It consisted of several tasks that required different approaches to element selection. In this paper we describe the approach that we adopted in an attempt to satisfy the requirements of all the tasks, Thorough, Focused, Relevant in Context, and Best in Context. We have used the same underlying system to approach all tasks. The retrieval strategy is based on the construction of a collection sub-tree, consisting of all nodes that contain one or more of the search terms. Nodes containing search terms were then assigned a score using the GPX ranking scheme which incorporates TF-IDF or BM25 variants, but extends them. Scores are recursively propagated to ancestors in the document XML tree, and finally all scoring XML elements are ranked. We present results that demonstrate that the approach is versatile and produces consistently good performance. We also provide empirical analysis of the GPX ranking scheme and compare its performance against a baseline TF-IDF and a BM25 scoring scheme.