Evaluation-Based Reasoning with Disjunctive Information in First-Order Knowledge Bases
Evaluation-Based Reasoning with Disjunctive Information in First-Order Knowledge Bases
复制标题
一阶知识库中基于析取信息的基于评价的推理
DOI：
--
复制
发表时间：
2002
期刊：
影响因子：
--
通讯作者：
G. Lakemeyer
中科院分区：
文献类型：
--
作者：
G. Lakemeyer
摘要 在之前的工作中，Levesque 提出了基于评估的推理过程，用于所谓的真知识库，相当于一组可能不完整、可能无限的无函数基础文字。该过程称为 V ；保持数据库查询评估的效率和逻辑健全性。此外，如果查询被限制为特殊的范式，则 V 在逻辑上也是完整的。在本文中，我们提出了对这项工作的扩展，以处理知识库中的分离信息。我们定义一个查询评估过程 X，它泛化 V 来处理相当于一组可能无限的无函数基础子句的 KB。在这种情况下，逻辑上合理且完整的演绎推理通常是不可判定的，因此，我们描述了我们愿意在 X 中放弃什么以保持易处理性。 1 引言 [8] 中认为，迄今为止唯一提出的在极大的一阶知识库（KB）上有效执行演绎推理的程序（涉及说
Abstract In previous work, Levesque proposed anevaluation-based reasoning procedure for so-called proper KBs, equivalent to a possibly in-complete possibly infinite set of function-freeground literals. The procedure, called V ; pre-served the efficiency and logical soundness ofdatabase query evaluation. Moreover, if thequery was constrained to be in a special normalform, V was also logically complete. In this pa-per, we propose an extension to this work to han-dle disjunctive information in a KB. We definea query evaluation procedure X that generalizes V to deal with KBs that are equivalent to a pos-sibly infinite set of function-free ground clauses.Deductive reasoning that is logically sound andcomplete in this case is undecidable in general,and so, we describe what we are willing to giveup in X to preserve tractability. 1 Introduction It was argued in [8] that the only procedure yet proposedto perform deductive reasoning effectively over extremelylarge first-order knowledge bases (KBs) (involving say