论文部分内容阅读
不精确推理和非单调推理是人工智能中两个重要的研究方面．分析了基于概率的不精确推理和基于真值维护系统的非单调推理的各自特点和局限性，将论据的优先级和可信度引入ＡＴＭＳ，提出了结合优先级和可信度的非单调真值维护系统，并给出其算法．它能有效地处理不完全领域知识下的推理，结合不精确推理和非单调推理的优点，克服两者的局限性．
Inaccurate reasoning and non-monotonic reasoning are two important aspects of artificial intelligence. The characteristics and limitations of inaccurate reasoning based on probability and nonmonotonic reasoning based on true value maintenance system are analyzed. The prioritization and credibility of arguments are introduced into ATMS, and a nonmonotonic truth model combining priority and credibility The value is maintained by the system and its algorithm is given. It can effectively deal with inference in incomplete domain knowledge, and overcome the limitations of both by combining the advantages of inaccurate reasoning and nonmonotonic reasoning.