Nonmonotonic Reasoning and Computational Knowledge Representation

非单调推理和计算知识表示

• 批准号: 0097278
• 负责人: Miroslaw Truszczynski
• 金额: $ 55万
• 依托单位: University of Kentucky Research Foundation
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2001
• 资助国家: 美国
• 起止时间: 2001-07-15 至 2005-06-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0097278&HistoricalAwards=false
• 关键词: Nonmonotonic; Reasoning; Computational; Knowledge; Representation

IIS-0097278Miroslaw Truszczynski, Raphael A. Finkel and Victor MarekUniversity of Kentucky$146,167 - 12 mosNonmonotonic Reasoning and Computational Knowledge RepresentationThis is the first year funding of a three year continuing award. The PI will study and implement computational knowledge-representation systems based on the paradigm of answer-set programming (ASP) with nonmonotonic logic that he recently identified, in order to demonstrate the practicality and effectiveness of the approach. Logic is most commonly used in knowledge representation as follows. To solve a problem we represent its constraints and the relevant background knowledge as a theory in the language of first-order logic (or its fragment). We formulate the goal (the statement of the problem) as a formula of the logic. We then use proof techniques to decide whether this formula follows from the theory. A proof of the formula, variable substitutions or both determine a solution. Taking a different approach, the PI will study and develop computational knowledge representation tools based on nonmonotonic logics rather than on the first-order logic. In addition, he departs from the single-intended model approach dominant in logic programming. Under the ASP paradigm, a theory in a nonmonotonic formalism is regarded as a specification of a family of sets - a collection of its intended models. Each model is viewed as a representation of a different single solution. The PI will investigate syntactic and semantic issues of ASP formalisms based on nonmonotonic logics, study methods for fast computing with these formalisms, develop practical implementations, and demonstrate effectiveness of answer-set programming engines and their applicability in knowledge representation. If successful, the work will establish answer-set programming as a viable approach to declarative programming, which in turn will provide AI researchers and practitioners with a new generation of computational tools for knowledge representation.

miroslaw Truszczynski, Raphael a . Finkel和Victor marek肯塔基大学146,167 - 12美元单调推理和计算知识表示这是一项为期三年的持续奖励的第一年资助。PI将研究和实现基于他最近确定的非单调逻辑的答案集编程（ASP）范式的计算知识表示系统，以证明该方法的实用性和有效性。逻辑在知识表示中最常用，如下所示。为了解决一个问题，我们用一阶逻辑（或其片段）的语言将其约束和相关背景知识表示为一个理论。我们把目标（问题的表述）表述为逻辑的公式。然后我们使用证明技术来确定这个公式是否可以从理论中推导出来。一个公式的证明，变量替换或两者都确定一个解。采用不同的方法，PI将研究和开发基于非单调逻辑而不是一阶逻辑的计算知识表示工具。此外，他背离了逻辑编程中占主导地位的单一目的模型方法。在ASP范式下，一个非单调形式的理论被视为一个集合族的规范——它的预期模型的集合。每个模型都被视为不同的单个解决方案的表示。PI将研究基于非单调逻辑的ASP形式的语法和语义问题，研究使用这些形式的快速计算方法，开发实际实现，并演示答案集编程引擎的有效性及其在知识表示中的适用性。如果成功，这项工作将建立答案集编程作为声明性编程的可行方法，这反过来将为人工智能研究人员和从业者提供新一代知识表示的计算工具。