Tools for Deductive Databases

Queries in deductive databases are expressed by pure Horn clauses following the principle of “say what you want and not how to get it.” Consequently, there is a heavier burden on the compiler (or interpreter) to find an efficient evaluation of a given query. The difficulty is compounded when constraints (e.g., <) are present in rules or queries, or when the database has object-oriented features, which introduce function symbols.

Research in the last decade has generated many ideas for efficient evaluation of queries. Gradually, these ideas have amalgamated into a smaller number of principles, but a complete coherence is still lacking. Some of these ideas and principles will be presented and compared while attempting to couch them in a coherent framework. In particular, the following will be discussed.

• Magic sets vs. envelopes.

• Query trees.

• Removing redundant rules and subgoals.

• Utilizing constraints most efficiently.

• Detecting termination and effective computability in the presence of function symbols.