Dislog - a system for reasoning in disjunctive deductive databases

D1sLOG - A System for Reasoning in Disjunctive Deductive Databases Dietmar Seipel, Helmut Thone University of Tübingen Sand 13, D - 72076 Tübingen, Germany { sei pel: thoene }@informatik. uni-t ue bingen. de Abstract D1sLoG is a system for reasoning in disjunctive -ded_uctive databases. It seeks to combine features of disjunctive logic programming, such as the sup port for incomplete information, with those of deductive databases, such as all-result inference capabilities. Severa} basic operators are provided for logi cal and non-monotonic reasoning: The logical consequence operator derives all 'Iogically implied disjunctive clauses from a disjunctive logic program. The non monotonic operators are semantically founded on generalizations of the well known closed-world-assumption. Reasoning in disjunctive deductive databases is very complex, even for small examples. Many different optimization tech niques are integrated in D1sLoG to speed up the application performance. The main techniques rely on a clause tree data structure allowing for an efficient and transparent evaluation_ '¡he operators of D1sLoG can be loaded from a library into a PROLOG application. D1s L0G itself is implemented as a meta-interpreter in SICSTUS-PR0L0G. 1 Introduction During the last couple of years, a growing interest in the field of disjunctive logic programming can be observed ( cf. workshops at the ICLP, ILPS and IFIP'94 confer ences ). A large number of theoretical publications have brought a deep insight into semantic issues and non-monotonic reasoning: Lobo, Minker, Rajasekar [18], Van Gelder, Ross, Schlipf [27] and Przymusinski [21) have developed alternative seman tics. Marek, Truszczynski [19] and Dix. [10] give an overview of non-monotonic rea soning. As a marriage of disjunctive logic programming and database technologies, disjunctive deductive databases can form the foundation far integrated high-end reasoning systems. Advanced applications in which incomplete information must be extensively analyzed are likely to benefit from this technology. Far instance, the analysis of independencies in probabilistic reasoning ( cf. Pearl [20] and Seipel, -325-