2 (南京大 学 计算机软件新技术国家重点实验室,江 苏 南 京 210093) A Preference Logic Based on.pdf

ISSN 1000-9825, CODEN RUXUEW E-mail: jos@ Journal of Software, Vol.18, No.11, November 2007, pp.2728?2739 DOI: 10.1360/jos182728 Tel/Fax: +86-10 ? 2007 by Journal of Software. All rights reserved. 一种基于多类型偏好的偏好逻辑 ? 张志政1,2+, 邢汉承1, 王蓁蓁1, 倪庆剑1 1(东南大学 计算机科学与工程学院,江苏 南京 210096) 2(南京大学 计算机软件新技术国家重点实验室,江苏 南京 210093) A Preference Logic Based on Various Kinds of Preferences ZHANG Zhi-Zheng1,2+, XING Han-Cheng1, WANG Zhen-Zhen1, NI Qing-Jian1 1(School of Computer Science and Engineering, Southeast University, Nanjing 210096, China) 2(State Key Laboratory for Novel Software Technology, Nanjing University, Nanjing 210093, China) + Corresponding author: Fax: +86-25 E-mail: seu_zzz@, Zhang ZZ, Xing HC, Wang ZZ, Ni QJ. A preference logic based on various kinds of preferences. Journal of Software, 2007,18(11):2728?2739. /1000-9825/18/2728.htm Abstract: Presently, a whole logic system is still absent to represent and reason various kinds of preferences. In this paper, a preference logic MPL (logic of many kinds of preference) is introduced to supply the gap. In addition, recurring to the minimal/maximal specificity principle, a non-monotonic semantics of MPL’s language LMPL is defined, and the application and capability of MPL are initially investigated by rewriting preferences expressed by ranked knowledge based into LMPL. Moreover, the conclusion and the prospective are presented in the end. Key words: preference; preference logic; preference strategy; ranked knowledge base 摘 要: 针对目前缺乏多类型偏好共存的偏好逻辑系统的现状,提出并构造了一个能够描述和推理多种类型偏好 的逻辑系统MPL(logic of many kinds of preference).在进一步提出MPL语言LMPL基于最粗糙/最细致描述原则的非单 调语义基础上,通过分级知识库这种常用偏好表示方法的LMPL重写,初步考察了LMPL表示能力,最后进行总结并提出 需要进一步研究解决的问题. 关键词: 偏好;偏好逻辑;偏好策略;分级知识库 中图法分类号: TP18 文献标识码: A 偏好是人工智能领域新的重要的研究内容.人工智能研究人员通常把偏好看作是Agent具有的一种内部状 态或者常识,以期采用偏好解释Agent的智能行为,实现快速、有效的偏好处理,并取得初步成果[1?3].偏好研究在 哲学逻辑中有长远的历史[4],特别是von Wright提出的偏好逻辑及其定义的强、弱和ceteris paribus这 3 类偏 好 [5],因其简单、直观的特点,对当前人工智能偏好的逻辑表示和推理研究具有直接和广