The Cheshire II system provides users with the ability to search using either natural language queries with probabilistic searching or conventional Boolean queries and term matching, as well as using both simultaneously. Although these are implemented within a single algorithm, they comprise two parallel logical search engines. This allows users to take advantage of the strengths of each search strategy and to create queries tailored to their own particular needs. Explicit Boolean query formulations are most useful in narrowly defined searches, or searches for a known item or known author. Alternately, natural language topic statements and probabilistic matching of queries to documents are more useful in subject searches when users rarely know the indexing vocabulary used to describe the sought for items.
Besides allowing users greater flexibility, the motivation for using two search methods follows from the observation that no single retrieval algorithm has been consistently proven to be better than any other algorithm for all types of searches. By combining the results of two retrieval algorithms, we can exploit the advantages and reduce the limitations of each system. In addition, in terms of probability ranking, the more evidence about the relationship between a query and a document that is available to the system, the more accurate the system will be in predicting the probability that the document will satisfy the user's need.