基于HNC理论的期刊文献知识元检索

摘要

This paper put forward a six-tuple representation of periodical literature knowledge element based on Hierarchical Network of Concepts (HNC) theory, and a new extraction method of the knowledge element based on dependency parsing and HNC theory after the analysis of text structure. It utilized the domain information expressed by expression form of HNC and the classifier built by maximum en-tropy model to classify and store the knowledge element. At last the semantic retrieval of knowledge element was realized through a two-stage retrieval.%针对期刊文献知识元提出了一种基于HNC理论的六元组表示方法，以及在文献结构分析基础上依据依存分析和HNC理论进行知识元的抽取。利用HNC符号的领域信息以及最大熵模型建立的分类器，对所抽取的知识元进行分类并存储，最后通过两阶段检索实现知识元的语义检索。