Word Pairs in Language Modeling for Information Retrieval

Previous language modeling approaches to information retrieval have focused primarily on single terms. The use of bigram models has been studied, but the restriction on word order and adjacency may not be justified for information retrieval. We propose a new language modeling approach to information retrieval that incorporates lexical affinities, or pairs of words that occur near each other, without a constraint on word order. The use of compound terms in the vector space model has been shown to outperform the vector model with only single terms (Nie & Dufort, 2002). We explore the use of compound terms in a language modeling approach, and compare our results with the vector space model, and unigram and bigram language model approaches.