常识知识表示:通向“人工智能”的途径

Commonsense Knowledge Representation:the Way to Artificial Intelligence

人工智能的知识表示包括专业知识与常识知识。人类常识知识具有语境相关性的显著特征,人工智能要处理常识知识就必须对与知识相关的语境进行研究。以麦卡锡、莱纳特为代表,两条不同路线的人工智能学者致力于常识知识的形式化研究并取得一些进展,但是当前人工智能知识表示中依然存在大量未解的难题。本文认为,寻找解决常识知识形式化这一人工智能研究的重要问题的关键一方面需要人工智能研究思路的转变,另一方面则在于从各门相关学科中寻找新的解题理念。

Human knowledge includes expert knowledge and commonsense knowledge. Artificial intelligence study must deal with commonsense knowledge and must deal with the context of human language. J. McCarthy and D. Lenet have done lots of formalizing context, but there are still problems left. Therefore, formalizing context is an important question in formalizing commonsense study, and this question is connected with the development of philosophy, logic and linguistics.