arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1804.08480cs.LO
收藏

基于极小模型和不可满足核的ASP中的谨慎推理

Cautious reasoning in ASP via minimal models and unsatisfiable cores

Mario Alviano, Carmine Dodaro, Matti Järvisalo, Marco Maratea, Alessandro Previti

摘要:

回答集编程(ASP)是一个基于逻辑的知识表示框架,支持-在其他推理模式中-查询回答的中心任务。在命题的情况下,查询回答相当于在给定的候选集合中的原子之间计算输入程序的谨慎后果,其中谨慎后果是属于所有稳定模型的原子。目前,最有效的算法要么迭代地验证一个候选人的补充扩展的输入程序的稳定模型的存在,其中候选人被选择,或引入一个子句强制执行至少一个候选人的虚假性,使得求解器可以自由选择在稳定模型的计算期间的任何时间伪造哪个候选人。本文介绍了计算谨慎后果的新算法,目的是驱动求解器搜索稳定的模型,放弃更多的候选人。具体来说,这样的算法之一强制极小的集合上的真实的候选人,其中可以使用不同的概念的极小,和另一个利用不可满足的核心计算。该算法在WASP中实现,并在最新的ASP竞赛的基准测试上进行实验,结果表明新算法的性能优于现有技术。(正在考虑接受TPLP)。

英文摘要:

Answer Set Programming (ASP) is a logic-based knowledge representation framework, supporting---among other reasoning modes---the central task of query answering. In the propositional case, query answering amounts to computing cautious consequences of the input program among the atoms in a given set of candidates, where a cautious consequence is an atom belonging to all stable models. Currently, the most efficient algorithms either iteratively verify the existence of a stable model of the input program extended with the complement of one candidate, where the candidate is heuristically selected, or introduce a clause enforcing the falsity of at least one candidate, so that the solver is free to choose which candidate to falsify at any time during the computation of a stable model. This paper introduces new algorithms for the computation of cautious consequences, with the aim of driving the solver to search for stable models discarding more candidates. Specifically, one of such algorithms enforces minimality on the set of true candidates, where different notions of minimality can be used, and another takes advantage of unsatisfiable cores computation. The algorithms are implemented in WASP, and experiments on benchmarks from the latest ASP competitions show that the new algorithms perform better than the state of the art. (Under consideration for acceptance in TPLP).

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑