首页> 外文学位 >Reasoning with uncertainty in deductive databases and logic programs
【24h】

Reasoning with uncertainty in deductive databases and logic programs

机译:演绎数据库和逻辑程序中的不确定性推理

获取原文
获取原文并翻译 | 示例

摘要

Of all scientific investigations into reasoning with uncertainty and chance, probability theory is perhaps the best understood paradigm. Nevertheless, all studies conducted thus far on the semantics of quantitative logic programming have been restricted to non-probabilistic semantical characterizations.;In this thesis, we develop a few frameworks to rectify this situation. In the first part of our study, we introduce a deductive database language which is expressive enough to represent such probabilistic relationships as conditional probabilities, Bayesian updates, probability propagation and mutual exclusion. We propose a fixpoint theory and a probabilistic model theory, and characterize their inter-relationships. Furthermore, we develop a proof procedure for this language, and present soundness and completeness results of this procedure.;As the aforementioned language is monotonic in nature, we discuss in the second half of this thesis three approaches of supporting non-monotonic probabilistic reasoning. In the first approach, we use a non-monotonic negation operator. We also present different semantical structures characterizing the semantics of such negations. All these structures are based on the stable semantics for classical logic programming. In the second approach, we use the Dempster-Shafer rule of combination. After developing a fixpoint theory for this approach, we present a precise relationship linking the Dempster-Shafer mode of non-monotonicity to the stable semantical approach introduced earlier.;So far, we have focussed entirely on using subjective probabilities. In the third approach, we discuss how empirical probabilities can be supported in monadic deductive databases. We first develop a model-theoretic characterization for such databases, and present a totally correct algorithm for checking consistency. Then we develop a sound and complete query answering procedure that supports non-monotonic reasoning based on changing reference classes.
机译:在所有关于不确定性和偶然性推理的科学研究中,概率论也许是最能被理解的范例。尽管如此,到目前为止,所有关于量化逻辑程序设计语义学的研究都局限于非概率性语义特征描述。本论文中,我们开发了一些框架来纠正这种情况。在我们的研究的第一部分中,我们引入了一种演绎性数据库语言,该语言足以表示条件概率,贝叶斯更新,概率传播和互斥等概率关系。我们提出了一个定点理论和一个概率模型理论,并描述了它们之间的相互关系。此外,我们开发了这种语言的证明程序,并给出了该程序的健全性和完整性结果。由于上述语言本质上是单调的,因此我们在本文的后半部分讨论了三种支持非单调概率推理的方法。在第一种方法中,我们使用非单调求反运算符。我们还提出了表征这种否定语义的不同语义结构。所有这些结构都基于经典逻辑编程的稳定语义。在第二种方法中,我们使用Dempster-Shafer组合规则。在为这种方法开发了一个定点理论之后,我们提出了一种精确的关系,将非单调性的Dempster-Shafer模式与之前介绍的稳定语义方法联系起来。到目前为止,我们完全集中在使用主观概率上。在第三种方法中,我们讨论了如何在单子演绎数据库中支持经验概率。我们首先为此类数据库开发模型理论特征,然后提出一种检查一致性的完全正确的算法。然后,我们开发一个完善且完整的查询回答过程,该过程支持基于更改的参考类的非单调推理。

著录项

  • 作者

    Ng, Raymond Tak-yan.;

  • 作者单位

    University of Maryland, College Park.;

  • 授予单位 University of Maryland, College Park.;
  • 学科 Computer science.
  • 学位 Ph.D.
  • 年度 1992
  • 页码 386 p.
  • 总页数 386
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类
  • 关键词

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号