York University at TREC 2006: Enterprise Email Discussion Search

Yu Fan, Xiangji Huang, Aijun An
2006 Text Retrieval Conference  
We use the Okapi retrieval system to conduct the email discussion search. The following issues are investigated. First, we make use of the thread structure in the emails to re-rank the documents retrieved by Okapi. We would like to see whether such post-processing of the retrieval result can boost the retrieval performance. Second, in terms of query formulation, we investigate whether the use of only title in a topic achieves better or worse results than the inclusion of other fields such as
more » ... cription and narrative. Third, we investigate whether stemming and stop word removal play an important role in the email search. Our conclusion includes that (1) re-ranking documents using a straightforward method that considers the thread structure can make a small improvement to the retrieval performance, (2) formulating the query using all the fields in a topic achieves the best result, and (3) the use of stemming and stop word removal can improve the performance, but the degree of improvement depends on the stemming method and the stop word list used.
dblp:conf/trec/FanHA06 fatcat:vviulfo7iven7f4dehcf365xj4