Patent US 6,594,658

The present mechanism relates to a method and apparatus for generating responses to queries to a document retrieval system. The system responds to a specific request for information by locating and ranking portions of text that may contain the information sought. It locates small relevant passages of text (called “hit passages”) and ranks them according to an estimate of the degree to which they correspond to the information sought. The system minimizes the number of these hit passages that need to be examined before an information seeker has either found the desired information or can safely conclude that the information sought is not in the collection of texts. A relaxation ranking mechanism is provided to accommodate paraphrase variations that occur between the description of the information sought and the content of the text passages that may constitute suitable answers, by retrieving phrases that are dissimilar to the query phrase to different degrees according to a predefined set of rules, and penalizing the retrieved phrases based upon the degree of this dissimilarity, thus providing the user with a priority organized query hit list.