Abstract: Knowledge acquisition is the key step in knowledge engineering, and acquiring professional knowledge from texts is an important and commonly used way. However, different texts have different description on the same object. In order to acquire complete knowledge of high-precise and fine-grained , this paper provides a method for acquiring history knowledge from multiple texts knowledge sources, which can be described as follows: automatically translating history knowledge into frames; inconsistency detection and revision, and merging related frames. The experimental results show that this method is feasible and effective. It can acquire high-precise and fine-grained history knowledge, and lay important basis for knowledge service.

Key words: frame representation, text knowledge acquisition, inconsistency detection, knowledge validation

摘要： 知识获取是知识工程中关键的一环，而从文本知识源中获取专业知识是一种重要而常用的途径。而不同文本知识源对同一对象描述往往有所不同，为了获取完备的高精度和细粒度的知识，给出了一种从多种文本知识源中获取历史知识的方法。其基本思想是：将不同文本知识源的知识自动翻译成框架知识；对描述相同对象的不同框架进行不一致性检测并修正；合并知识框架。实验表明，该方法能获取到高精度和细粒度的历史知识，是可行和有效的，并为下一步的知识服务奠定重要的基础。

关键词: 框架表示法, 文本知识获取, 不一致性检测, 知识修正