Term relevance estimation for Chinese query expansion

In this paper we propose a novel method to estimate the relevance between query and candidate expansion terms for Chinese information retrieval. In previous method, expansion terms are usually selected by counting term co-occurrences in the documents. However, term co-occurrences are not always a good indicator for relevance, whereas some are background terms of the whole collection. In order to remove noise, an EM-algorithm is used in our model to estimate two kinds of relevance weight. One is the relevance weight between query and its relevant term extracted from the top-ranked documents in initial retrieval results. The other is the relevance weight between each query term and its relevant terms extracted from the snapshot of Google search result when that query term is used as search keyword. The estimated relevance weights are used to select good expansion terms for second retrieval. The experiments on the two test collections show that our query expansion model is more effective than the standard Rocchio expansion.