基于演化Agent的推理模型  

A REASONING MODEL BASED ON EVOLUTING AGENT

作　　者：印鉴[1] 汤庸[2] 

机构地区：[1]中山大学计算机科学系,广州510275 [2]广东工业大学计算机科学系,广州510090

出　　处：《小型微型计算机系统》2000年第3期278-281,共4页Journal of Chinese Computer Systems

基　　金：国家自然科学基金;广东省自然科学基金

摘　　要：本文描述了一种基于演化 agent的推理模型 ,并用这种推理模型来处理定理机器证明 .传统上 ,定理机器证明常常使用某种逻辑表示 ,然后再进行推理 ,这些方法往往缺乏灵活性 ,且证明过程难以理解 .在本文所叙述的方法中 ,演化 agent能将目标即待证定理分解成越来越小且越来越容易证明的子目标 ,最后完成定理证明 .这种方法非常类似于人类在证明定理时一般所采用的思维方式 ,因而 ,显得更灵活。This paper presents a reasoning model based on evoluting agent and used it to deal with the problem of mechanical theorem proving. Traditionally, Mechanical theorem proving often use a logic representation , and then inference, these methods lack of flexibility and the proving processes are difficult to understand. In the proposed approach, the evoluting agent can mutate the goal the theorem to be proved into subgoals, these subgoals can easily be proved, thus, in this way, a goal can gradually be mutated into some easily proved proposition or fact. The processes of this method are similar to the human beings, and it is much more adaptive, flexible and creative.

关 键 词：AGENT 定理机器证明 人工智能 推理模型 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]