基于中文维基百科的领域概念相关性研究

Research on Semantic Relatedness of Domain-specific Concepts Based on Chinese Wikipedia

以提高领域概念相关性判断的准确度为研究宗旨,提出综合利用中文维基百科的分类体系结构和概念释义内容进行概念间语义相关度计算的方法。选取中文维基百科分类体系下的图书情报领域的概念为实验对象,将基于分类信息和文本信息的加权算法与单独基于分类信息的语义距离算法和信息量算法,以及基于文本信息的文本重叠算法进行对比分析。实验结果表明:加权算法能取得更好的效果,可为实现面向领域的信息检索、领域本体构建等应用提供重要技术支持。

In order to improve the accuracy of computing the relatedness of the domain-specific concepts, this paper proposes a new semantic relatedness algorithm using Chinese Wikipedia category architecture and concept interpretation content. The concepts in library and information science in concept-hierarchy of Chinese Wikipedia are taken as experiment objects, and weighted algorithm based on category and text information are compared with other algorithms only based on Chinese Wikipedia category like Relsctup and Rel or on Chinese Wikipedia article like Rel,r. The experimental results show that the weighted algorithm is better than the others, and provide important technical support for application such as domain - oriented information retrieval, construction of domain ontology and so on.