Using text classification method in relevance feedback

Abstract

In modern Information Retrieval, traditional relevance feedback techniques, which utilize the terms in the relevant documents to enrich the user's initial query, is an effective method to improve retrieval performance. In this paper, we re-examine this method and show that it does not hold in reality - many expansion terms identified in traditional approaches are indeed unrelated to the query and harmful to the retrieval. We then propose a Text Classification Based method for relevance feedback. The classifier trained on the feedback documents can classify the rest of the documents. Thus, in the result list, the relevant documents will be in front of the non-relevant documents. This new approach avoids modifying the query via text classification algorithm in the relevance feedback, and it is a new direction for the relevance feedback techniques. Our Experiments on TREC dataset demonstrate that retrieval effectiveness can be much improved when text classification is used. © 2010 Springer-Verlag Berlin Heidelberg.