This book is based on the doctoral dissertation of R. Al-Asady, completed at Exeter University, in 1993. It is concerned with creating an inheritance theroy from the AI point of view. Within Artificial Intelligence, the need to create sophisticated intelligent behavior based on commonsense reasoning has long been recognized. Such commonsense reasoning is characterized by the withdrawing of previously drawn conclusions when new information comes along. Research has demonstrated theft formalisms for dealing with commonsense reasoning require nonmonotonic capabilities where, typically, inferences based on incomplete knowledge need to be revised in the light of later information that fills in some of the gaps. In this book, an inheritance theory based on multiple inheritance structures with exceptions (nonmonotonic inheritance structures) is proposed. Without an adequate nonmonotonic inheritance reasoning technique, such as exceptional inheritance reasoning (EIR) as proposed in this volume, inheritance networks will produce inconsistencies. A number of nonmonotonic properties that enable EIR to subsume existing formalisms such as default logic and inferential distance ordering have been included within the reasoning technique presented here. An inheritance algorithm is also presented and a demonstration is included to show how it can be used to specify and implement various nonmonotonic inheritance problems. In addition, an inheritance formalism has been developed that is capable of dealing with ambiguous situations and can handle other classes of nonmonotonic problems apart from those already presented in the literature.

Finally, in the application section, this inheritance formalism has been applied to two important domains, namely causal reasoning and analogical reasoning, to demonstrate the conceptual power and expressiveness of the formalism.

本书是基于R.Al-Asady的博士论文，于1993年在埃克塞特大学完成。它涉及到从人工智能的角度创建一个继承理论。在人工智能领域，人们早已认识到需要在常识性推理的基础上创造复杂的智能行为。这种常识性推理的特点是，当新的信息出现时，会撤回之前得出的结论。研究表明，处理常识性推理的形式化需要非单调性的能力，通常情况下，基于不完整知识的推论需要根据后来的信息来修正，以填补一些空白。在这本书中，我们提出了一个基于有例外的多重继承结构（非单调继承结构）的继承理论。如果没有适当的非单调性继承推理技术，如本卷提出的例外继承推理（EIR），继承网络将产生不一致。这里提出的推理技术中包含了一些非单调性的属性，这些属性使EIR能够归入现有的形式主义，如默认逻辑和推理距离排序。还介绍了一种继承算法，并包括一个演示，以说明如何使用它来指定和实现各种非单调的继承问题。此外，我们还开发了一个继承形式主义，它能够处理模棱两可的情况，并能处理除文献中已经提出的问题之外的其他类别的非单调问题。

最后，在应用部分，该继承形式主义被应用于两个重要领域，即因果推理和类比推理，以证明该形式主义的概念能力和表现力。