中文领域知识半自动化 OWL 本体构建方法研究

摘要

Traditional Chinese domain ontology constructions are usually by manual,such method is time-consuming and lower in reuse rate.Aiming at the problem,we put forward a method that can semi-automatically convert Chinese domain knowledge documents into OWL ontology.Taking weapon equipment domain as an example,first the method used the word segmentation tool ICTCLAS5.0 to carry out pretreatment of segmentation on Chinese documents,then used TF-IDF algorithm and the template-based matching algorithm to mine the core concepts and clustering concept hierarchy relationship,and after that it formalised the clustered triple documents to OWL ontology using Jena, and finally conducted visualisation management on OWL ontology by Protégétools.Experimental results showed that,this method could effectively save the time and manpower costs of Chinese ontology construction,and provided a new thought and attempt to fully solving the automatic construction of Chinese domain ontology.%传统中文领域本体构建多采用手工构建的方法，这种方法费时费力且重用率不高。针对这种情况，提出一种把中文领域知识文档半自动化地转化为 OWL（Web Ontology Language）本体的方法。该方法以武器装备领域为例，先使用分词工具 ICTCLAS5．0对中文文档进行分词等预处理，然后使用 TF-IDF 算法、基于模板匹配算法挖掘领域核心概念、聚类概念等级关系，再使用 Jena 将聚类后的三元组文档形式化为 OWL 本体，最后使用 Protégé工具对 OWL 本体进行可视化管理。实验结果表明，该方法可以有效地节省中文本体构建的时间和人力成本，为全面解决中文领域本体的自动化构建提供了一个新的思路和尝试。