We present the results of our enterprise expert search system application to the tasks that were introduced at the Text Retrieval Conference (TREC) in 2005—2007. The expert search system is based on the analysis of content and communications topology in an enterprise information space. During the performed experiments an optimal set of weighting coefficients for three query-candidate associating algorithms is selected for achieving the best search efficiency on a specified corpus. The obtained performance proved to be better than at most TREC participants. The hypothesis of additional efficiency improvement by means of query classification is proposed.