Ricoh Research at TREC 2006: Enterprise Track
نویسندگان
چکیده
This article presents our contributions to expert search and discussion search of Enterprise Track in TREC 2006. In discussion search, we take advantage of the redundant patterns of emails, such as the subject, author, sent time, etc., which we incorporate in a field-based weighting method to mine discussion topics with more robustness. Some non-content features, such as time-line and mail thread are found to be useful as experiments showed they improve the precision of the search. In expert search, two variants of the BM25 and DFR_BM25 weighting models namely V-BM25 and V-DFR_BM25 are put forward. Query-based document length, not profile length, is used as document length in these weighting models to eliminate multiple topic drift. In addition, we propose a variant of an existing phrase weighting model to decrease topic confusion (V-phrase) and a two-stage field-based search method to refine the results. We demonstrate these approaches can effectively improve expert search.
منابع مشابه
TREC 2006 at Maryland: Blog, Enterprise, Legal and QA Tracks
In TREC 2006, teams from the University of Maryland participated in the Blog track, the Expert Search task of the Enterprise track, the Complex Interactive Question Answering task of the Question Answering track, and the Legal track. This paper reports our results.
متن کاملUMass at TREC 2006: Enterprise Track
This paper gives an overview of the work done at the University of Massachusetts, Amherst for the TREC 2006 Enterprise track. For the discussion search task, we compare two methods for incorporating thread evidence into the language models of email messages. For the expert finding task, we create implicit expert representations as mixtures of language models from associated documents.
متن کاملL3S Research Center at TREC 2006 Enterprise Track
The L3S Research Center submitted four runs at Enterprise Track for the first time in 2006, all of them are based solely on the W3C mailing lists. The first run serves as a fully automatically produced baseline. The second run uses a threshold on the document scores to limit the number of documents used for expert ranking. The third uses in addition a threshold on the experts scores in order to...
متن کاملDUTIR at TREC 2006: Genomics and Enterprise Tracks
This paper describes the techniques we applied for the two TREC 2006 tracks, i.e., Genomics and Enterprise track. For the Genomics Track, we used a Rocchio relevance feedback method to expand the terms and then performed passage retrieval by building dual index and using half overlapped windows passages. Several approaches to merge the results and rerank the passages are presented. For the Ente...
متن کامل