Home DLP*
Download

Back to MAG
A Disjunctive Logic Program (DLP) is a collection of logical rules where disjunction is allowed in the heads and negation may occur in the bodies. The semantics of DLP is based on the notion of answer set; a DLP may have several alternative answer sets (possibly none), each corresponding to a possible view of the world. DLP under the answer set semantics, often referred to as Answer Set Programming (ASP), evolved significantly during the last decade, and has been recognized as a convenient and powerful method for declarative knowledge representation and reasoning. Several systems supporting ASP have been implemented, thereby encouraging a number of applications in several real-world contexts ranging, e.g., from information integration, to frauds detection, to software configuration. On the one hand, these applications have confirmed the advantages of DLP for knowledge-based tasks. On the other hand, they evidenced some serious drawbacks of both the language and the present DLP reasoning engines. The aim of this project is to move a first step to overcome these deficiencies. The overall project goal is the achievement of a powerful extension of DLP, which can be profitably used in advanced real-world applications. Specifically, we will first enrich the DLP language to support typing and complex terms, like functions, sets, and lists, which are useful for reasoning about recursive data structures and infinite domains (as required, e.g., while reasoning on XML/HTML documents). Then, we will investigate suitable mechanisms for interfacing DLP with objectoriented languages (in order to exploit DLP for the 'intelligent' modules, while implementing the whole infrastructure in imperative languages). Finally, we will design heuristics and advanced optimization techniques, to enhance DLP systems and make them able to deal also with huge amounts of data. Contemporarily, the applicability of the new language in various interesting applications, such as automated reasoning on Web data and validation of Objectoriented specifications, will be investigated. In the first phase of the project, the theoretical underpinnings of the above-described extensions will be investigated, and a DLP extension, called DLP*, incorporating them will be proposed. Several interesting problems are to be faced. For instance, inference in DLP enriched with function symbols is highly undecidable, if no restriction is imposed. Therefore, we will single out decidable fragments and design evaluation algorithms for reasoning in such fragments. In this phase, the unit II (UNINA) will give his main contribution, because of its high-level expertise in theoretical and algorithmic problems related to the introduction of functions in DLP. In this phase, the unit I (UNICAL), in cooperation with the unit III (UNIME), will single out the needed linguistic extensions, and will provide a formal definition of both their syntax and semantics. In these activities, UNICAL can fruitfully exploit its wide experience on logic-based languages for knowledge representation. In the second phase, a prototype supporting DLP* will be implemented. This prototype will allow to test the effectiveness of the proposed techniques, and check the usefulness of the proposed extensions in practice. To this end, UNICAL will develop the prototype as an extension of the state-of-the art DLP system DLV. For this activity, UNICAL will exploit its indepth expertise on the development of DLP systems, which has achieved during the development of DLV (in collaboration with the Technical University of Vienna). In parallel with these two phases, UNIME will study the applicability of the new language to complex and challenging application domains by formally specifying their requisites; in turn, this information will be exploited to better specify the features to be included in the language extensions. Finally, UNIME and UNINA will take care of testing, validation, and assessment of the developed system.
Eurolinux Software Patent Petition Blue Ribbon Campaign PHP Language Concurrent Versioning System SubVersion Valid HTML 4.01! Strict Valid CSS!