4 Conclusions and Future Directions Example 5 Existential Query Example 6 Intelligent Backtracking. 3 Compilation and Execution Example 4 Rule Compression and Literal Collapsing for Aggregate Operators Example 3 Built-in Aggregate Operators Example 2 Sql Generation

A. Van Gelder,K. A. Ross,Jayen Vaghani,Kotagiri Ramamohanarao,David B. Kemp,Zoltan Somogyi,Peter J. Stuckey,Tim S. Leask,James Harland

2007-01-01

Abstract:The well-founded semantics for general logic programs. 11 Therefore using choice and stratiied negation it is possible to express all algorithms that can be computed in time which is polynomial in the size of the database 7]. Now, while the declarative semantics of choice rests on the theoretical foundations of non-monotonic logic, a programmer need not be cognizant of the notion of stable models to make eeective use of choice in his/her program; indeed, there is an equivalent operational semantics based on the notion of memorizing the old values of choice and checking every candidate new value for violations of the FD constraints. This operational semantics is also the basis for eecient implementations of the construct. Likewise, the notion of XY-stratiication 21] a combination of formal non-monotonic semantics, intuitive appeal and amenability to eecient implementations are the ingredients that allow the relaxation of the stratiication requirement in LDL++. The realization of the LDL++ prototype, a second-generation system, demonstrates the signiicant progress recently made by deductive databases. As a result of this progress, the eld that was still in its infancy in the 80's can now claim (i) an implementation technology as advanced as that of relational databases, logic programming and AI expert systems, and (ii) theoretical foundations that combine and extend those of the before-mentioned elds is several domains, particularly non-monotonic reasoning. Furthermore, the deployment of the LDL++ prototype in various applications has much contributed to proving the unique advantages ooered by deductive databases in several domains 18, 19, 12]. Future evolution of the eld could be shaped by various advances that have taken place after the completion of the LDL++ prototype. Indeed, recent progress makes it possible to have a system that integrates active database rules with deductive rules, and supports in a uniied framework temporal reasoning, spatial reasoning and reasoning in the presence of uncertainty. Furthermore, important application areas are now emerging for this technology, including middleware systems 11] and knowledge discovery 16]. Ideally, this combination of application pull and technology push will soon produce a new-generation of deductive database systems that take us beyond the current LDL++ and its second-generation technology. 10 such as choice((X); (Y)) is used to denote that the functional dependency X ! Y must hold in the model deening the meaning of this program. For instance the following program, assign to each student an advisor from the same area: Thus, while the result …