On program grounding in ASP

Baselice, Sabrina (2008) On program grounding in ASP. [Tesi di dottorato] (Inedito)

Full text disponibile come:

[img]PDF - Richiede un editor Pdf del tipo GSview, Xpdf o Adobe Acrobat Reader
599Kb

Abstract

Answer set programming (ASP) is a declarative problem solving framework introduced by Michael Gelfond and Vladimir Lifschitz in the late ’80s. ASP has received much attention by researchers for its expressiveness and simpleness so that well-engineered and optimized implementations have been developed for it. However, state-of-the-art answer set solvers have still a strong limitation: they are not be able to reason on nonground programs and then the input program have to be instantiated before the solver can start to reason on it. Consequently, answer set solvers (i) cannot handle infinite domains and (ii) use huge amounts of memory even if domains are finite. This work wants to give some contribution for these two not trivial problems. First, I analyze finitary programs as a class of programs that can effectively deal with function symbols and recursion (hence infinite domains and models). Interestingly, even if finitary programs are computationally complete, their restrictions make it possible to keep complexity under control. I study the consequences of relaxing the restrictions on finitary programs and my results enforce a kind of minimality of the properties that characterize finitary programs. Next, I investigate what happens when we “compose” two programs P and Q belonging to some particular classes that imposing them some restrictions guarantee good computational properties, so obtaining a program P [ Q that, as a whole, might not be subject to the restrictions of P or Q but that again enjoys good computational properties. Finally, I study a new approach to tackle the problem (ii) of ASP. The idea is to integrate answer set generation and constraint solving to reduce the memory requirements for a class of multi-sorted logic programs with cardinality constraints: constrained programs. I prove some theoretical results, introduce provably sound and complete algorithms, and report experimental results on my prototype system for evaluating constrained programs, showing that my approach can solve problem instances with significantly larger domains.

Tipologia di documento:Tesi di dottorato
Parole chiave:Knowledge representation and reasoning
Settori scientifico-disciplinari MIUR:Area 01 Scienze matematiche e informatiche > INF/01 INFORMATICA
Coordinatori della Scuola di dottorato:
Coordinatore del Corso di dottoratoe-mail (se nota)
De Luca, Aldo
Tutor della Scuola di dottorato:
Tutor del Corso di dottoratoe-mail (se nota)
Bonatti, Piero Andrea
Stato del full text:Accessibile
Data:2008
Numero di pagine:125
Istituzione:Università degli Studi di Napoli Federico II
Dipartimento o Struttura:Matematica ed Applicazioni
Tipo di tesi:Dottorato
Stato dell'Eprint:Inedito
Denominazione del dottorato:Scienze Computazionali ed Informatiche
Ciclo di dottorato:XX
Numero di sistema:1992
Depositato il:29 Luglio 2008
Ultima modifica:04 Febbraio 2009 09:48

Solo per gli Amministratori dell'archivio: edita il record