Using SQL with Prolog to improve performance with large databases

This article describes how the performance of certain Prolog programs can be improved by storing large lists of facts in an SQL database rather than as Prolog facts. In experiments that will be described, the speed improvements ranged from negligible to a factor of over 200. This improvement comes about because SQL servers are strongly optimized for searching large, flat tables. However, modern Prolog compilers are increasingly good at handling large volumes of data, and not all applications will benefit from the use of SQL. This article suggests which types of application are likely to benefit, and discusses other ways in which the use of SQL may be of benefit in Prolog programming.