Towards expressive KR systems integrating datalog and description logics: preliminary report

We study the problem of extending hybrid knowledge-based systems integrating rules and structured representation of information. Speciically, we deene disjunctive AL-log, a hybrid formalism which combines two very expressive knowledge representation languages: disjunctive datalog and the description logic ALC. First, by exploiting the use of nonmono-tonic features of disjunctive datalog, we show that disjunctive AL-log overcomes some severe representational limitations of analogous proposals. Then, in order to implement a system based on disjunctive AL-log, we take advantage of recent advances in the technology of KR systems: in particular, we are able to deene a reasoning method that allows for a full implementation of disjunctive AL-log, obtained by relying on the optimized deductive services of two state-of-the-art implemented KR systems, i.e. the DL-based system fact and the disjunc-tive datalog system dlv. We also provide both a computational characterization of reasoning in disjunctive AL-log and a general technique for encoding extensional knowledge (ABox) in the system fact.