The Alexander Method — A technique for the processing of recursive axioms in deductive databases

We propose a technique for handling recursive axioms in deductive databases. More precisely, we solve the following problem:Given a relational query including virtual relations defined from axioms (Horn clauses, with variables in the conclusion predefined in the hypotheses), which can be recursive, how to translate this query into arelational program, i. e. a set of relational operations concerning only real relations (not virtual). Our solution has the following properties:• the program to evaluate the query always terminates,• the relational program is produced by a pure compilation of a source query and of the axioms, and is independent of the data values (there is no run-time),• the relational operations are optimized: theyfocus towards the computation of the query, without needless computations.As far as we know, the Alexander Method is the first solution exhibiting all these properties. This work is partly funded by Esprit Project 112 (KIMS).