基于语义相似度的本体概念更新方法研究

摘要

To realize the automatic updating of ontology concepts and reduce the excessive dependence on domain experts,a semantic similarity-based ontology concept update method (SSOCUM) was proposed.An improved similarity algorithm based on WordNet was implemented.Based on the calculation of the path length,the algorithm took into account the concept of the node depth and the impact of the amount of information on the similarity.To make up several inadequacies that semantic information of property was neglected by WordNet-based similarity algorithm,property similarity was taken into account to make an adjustment.Through experimental comparison,it is verified that the Pearson coefficient between the improved algorithm and the standard data set is higher than the traditional algorithm,and the calculation result as closer to human subjective judgment.The constructed SSOCUM algorithm was used to analyse the ontology of coal mine ventilation system.The results show that the SSOCUM algorithm can help to automatically add ontology new concepts,and has certain accuracy and validity.%为了实现本体概念的自动更新,减少对领域专家的过多依赖,给出一种基于语义相似度的本体概念更新方法SSOCUM(Semantic Similarity-based Ontology Concept Update Method).实现一种改进的基于WordNet的相似度算法,该算法在计算路径长度的基础上,综合考虑了概念的节点深度以及信息量对相似度的影响.为了弥补基于WordNet的相似度算法没有考虑概念属性所携带的语义信息的不足,加入属性相似度对其进行调整.通过实验对比,验证了改进算法的计算结果与标准数据集之间的皮尔森系数高于传统算法,计算结果更接近于人的主观判断.采用构建好的煤矿领域通风系统本体对SSOCUM算法进行实验分析.结果表明,SSOCUM算法有助于本体新概念的自动添加,并具有一定的准确性和有效性.