基于FCA的领域问句相似度计算方法

Calculation method of domain question similarity based on FCA

文章提出一种基于形式概念分析的领域问句相似度计算方法,该方法不仅考虑了问句的语义与语法结构,更主要的是利用形式概念分析方法提取和建立领域概念集,是一种新的计算领域问句相似度计算模型。通过形式概念分析构建领域知识概念集,概念集以概念格形式进行组织。将问句相似度计算转换成概念集的相似度计算。试验结果表明,使用该方法的相似度计算准确率约为84%,并在有噪声的情况下依然保持高准确率。

A domain question similarity calculation method was proposed based on formal concept analysis. This method not only considers the question of semantic and grammatical structure, but also uses the method of formal concept analysis to extract and build domain concept sets. It＇s a new model for computing the similarity of calculation questions. Concept sets of domain knowledge are built through formal concept analysis, and the concept sets are organized in the form of conceptual lattice. And then the similarity of questions can be transformed into the similarity of concept sets. Finally, the experiment results indicated that the accuracy of similarity calculation by using this method was about 84%, and it could maintain a high accuracy in the case of noise.