Efficiency Considerations on Built-in Taxonomic Reasoning in PROLOG

The issue of integrating inheritance rules into Prolog has recently received some attention in l i te ra ture. A Prolog-based interpreter extended with a special, bu i l t i n mechanism for handling is_a taxonomies has been bu i l t . The motivation for this work stems from two observations: (1) is_a hierarchies are common in many domains, and (2) the issue of representing in Prolog is_a hierarchies, and the inheritance properties related to i t , is not so straightforward as it would seem at a f i r s t glance, especially whether time and space ef f ic ien­ cy are required. The model underlying the proposed extension is shown and compared with relevant l i te ra ture. We describe how some new capabil i t ies have been added to a standard Prolog interpreter in order to implement the extended interpreter. In part icular, the unif icat ion algorithm and the management of some data areas have been modified. The space requirements and time performance of the extended interpreter are compared with those of the original standard Prolog interpreter, and the results of a series of tests are discussed.