Heuristics to locate the best document set in information retrieval systems

The use of best-match search strategies in information retrieval systems is discussed. In response to a given query, best-match searching requires the identification of those documents in the collection which are most similar to the query, with similarity being measured by an appropriate closeness function. The emphasis is on heuristics to efficiently locate the closest documents set. The problem is introduced with reference to a straightforward search procedure that returns the best documents manipulating inverted index entries. An improved algorithm is presented which computes in advance an upper bound on closeness, avoiding the exact computation of closeness in many instances and thus optimizing both the number of documents to be evaluated and the number of inverted lists to be inspected. The algorithm is analyzed, and experimental results are reported.<>