基于可信度的辩论模型及争议评价算法

辩论是智能主体间为了消除分歧的一种基于言语的交互行为.由于知识的局限性，争议以及争议内部的陈述通常存在不确定性，因此在对辩论进行建模时需要考虑不确定信息处理问题.提出一种基于可信度的辩论模型（CFA），该模型将争议表示为由若干前提和一个结论组成的可废止规则，并用对话树描述辩论推演过程.为了表示不确定性推理，引入可信度模型，将争议前提的不确定性和争议之间的攻击强度统一用可信度因子表示.在此基础上，提出计算陈述可信度的争议评价算法，并通过设定可信度阈值确定陈述的可接受性，得出最终辩论结果.最后，用一个实例说明该方法的有效性.该模型可以有效处理不确定信息条件下辩论推理过程，其辩论算法建立在数值计算基础之上，所得出的可接受陈述集在给定可信度阈值条件下是唯一的，可以克服Dung 的抽象辩论框架中扩充语义的不足.

Argumentation Model Based on Certainty-Factor and Algorithms of Argument Evaluation

Argumentation is a kind of interaction between agents for resolving differences in speech. Due to the limitation of knowledge, arguments and its inner statements are usually uncertain, and therefore it is necessary to consider uncertain information processing when modeling the process of argumentation. This paper proposes an argumentation model based on the theory of certainty-factor, in which an argument is expressed as a defeasible rule with several premises but only one conclusion, and the process of argumentation is described as a dialogue tree. In order to do uncertain reasoning, the uncertainty of argument's premise and the strength of the attack that one argument makes against another are assigned with certainty-factors. Next, the algorithms of argument evaluation are provided to compute the credibility of statements. Using the algorithms the set of acceptable arguments can be determined by the credibility threshold and the result of argumentation can be obtained. Finally, an example is given to illustrate the validity of the method. The algorithms are based on numerical computation where the set of acceptable arguments is unique under a given credibility threshold, and can overcome the shortage of extension semantics of Dung's abstract argumentation framework.