CLARE-a CLAuse Retrieval Engine for large Prolog databases

Although the logic programming language Prolog was originally designed to handle AI applications, its suitability for implementing database applications has been recognised for some while now. Its attractiveness for this purpose lies in the additional benefits which it can provide. It enables database systems to be developed which are considerably more powerful than current relational databases-systems which can store, retrieve and manipulate knowledge in the form of logical rules as well as conventional data in the form of facts. Such systems are generally referred to as deductive or logic database systems. However, existing Prolog systems are limited in their capacity to deal with large volumes of data. The basic unit of information in Prolog is the clause. Each clause may be a rule or a fact (corresponding to a tuple in a relational database). The process of matching a Prolog query against a clause in the database, termed unification, is much more complex than the simple associative matching performed in relational database systems, and consumes a considerable proportion of the execution time of a Prolog process. For a large database stored on disc, the time taken to attempt to unify a query against each clause in the database in turn would be intolerable. The authors describe a special purpose engine which has been designed to overcome this problem