Word Distribution Analysis for Relevance Ranking and Query Expansion

Apart from the frequency of terms in a document collection, the distribution of words plays an important role in determining the relevance of documents for a given search query. In this paper, word distribution analysis as a novel approach for using descriptive statistics to calculate a compressed representation of word positions in a document corpus is introduced. Based on this statistical approximation, two methods for improving the evaluation of document relevance are proposed: (a) a relevance ranking procedure based on how query terms are distributed over initially retrieved documents, and (b) a query expansion technique based on overlapping the distributions of terms in the top-ranked documents. Experimental results obtained for the TREC-8 document collection demonstrate that the proposed approach leads to an improvement of about 6.6% over the term frequency/inverse document frequency weighting scheme without applying query reformulation or relevance feedback techniques.