The University College London at TREC 2008 Enterprise Track

The University College London Information Retrieval Group participated in both the Expert Search and Document Search tasks in the TREC2008 Enterprise Track. We used a generic two-stage approach, which consists of a document retrieval stage followed by an expert association discovery stage, for expert finding. Since document search is an integral part of our expert finding approach, we have studied the relationship between document search and expert search. Due to the existence of rich features that can potentially contribute to expert finding, our expert finding approach integrates these features including anchor texts, indegree, and multiple levels of associations between experts and query terms. Our experimental results show that the introduction of features has helped improve the expert finding performance.