Ein Indexierungskonzept für LISPLOG-Datenbasen

Abstract

LISPLOG is a LISP/PROLOG integration explored at the Universitaet Kaiserslautern. To improve the performance of the LISPLOG interpreter, an indexing concept has been developed and implemented. Higher efficiency is gained by eliminating unnecessary unifications. Matching clauses are selected using the predicate and one constant atomic argument of the conclusion. A special combination of shared lists and binary trees is maintained by the indexing procedure to make this selection rather fast. The user may optionally specify the indexing argument; the default-argument is changeable when installing the system.

The indexing concept and the implemented algorithms and data structures are fully described in this paper, together with a complete source-listing (FRANZ LISP) of the implementation.