Techniques for efficient lazy-grounding ASP solving

Tutkimustuotos: Artikkeli kirjassa/konferenssijulkaisussavertaisarvioitu

Tutkijat

  • Lorenz Leutgeb
  • Antonius Weinzierl

Organisaatiot

  • Vienna University of Technology

Kuvaus

Answer-Set Programming (ASP) is a well-known and expressive logic programming paradigm based on efficient solvers. State-of-the-art ASP solvers require the ASP program to be variable-free, they thus ground the program upfront at the cost of a potential exponential explosion of the space required. Lazy-grounding, where solving and grounding are interleaved, circumvents this grounding bottleneck, but the resulting solvers lack many important search techniques and optimizations. The recently introduced ASP solver Alpha combines lazy-grounding with conflict-driven nogood learning (CDNL), a core technique of efficient ASP solving. This work presents how techniques for efficient propagation can be lifted to the lazy-grounding setting. The Alpha solver and its components are presented and detailed benchmarks comparing Alpha to other ASP solvers demonstrate the feasibility of this approach.

Yksityiskohdat

AlkuperäiskieliEnglanti
OtsikkoDeclarative Programming and Knowledge Management - Conference on Declarative Programming, DECLARE 2017, Unifying INAP, WFLP, and WLP, 2017, Revised Selected Papers
TilaJulkaistu - 1 tammikuuta 2018
OKM-julkaisutyyppiA4 Artikkeli konferenssijulkaisuussa
TapahtumaInternational Conference on Declarative Programming - Wurzburg, Saksa
Kesto: 19 syyskuuta 201722 syyskuuta 2017

Julkaisusarja

NimiLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
KustantajaSpringer
Vuosikerta10997 LNAI
ISSN (painettu)0302-9743
ISSN (elektroninen)1611-3349

Conference

ConferenceInternational Conference on Declarative Programming
LyhennettäDECLARE
MaaSaksa
KaupunkiWurzburg
Ajanjakso19/09/201722/09/2017

ID: 28944955