摘要
基于图理论的概念间语义度量方法,改进了语义相似度部分影响因素,提出一种结合设计良好的领域本体来计算自然语言概念间的语义相似度的算法。对自然语言与本体的关系进行分析,并通过本体对节点密度、节点深度与节点层次顺序等影响概念语义相似度的因素进行了改进,综合考虑概念的语义距离、概念间关系、概念的属性与概念所处的层次等影响因素,利用本体对相关领域的基本术语和关系的准确定义,改进了基于本体的概念间语义相似度的算法。实验结果表明,该算法对于提高概念间相似度的计算精度明显高于其他算法。
Based on graph theory similarity are improved. A natural concept semantic measurement method, the influencing factors of semantic language concept semantics similarity algorithm is presented. The relationships between natural language and ontology are analyzed, and the factors affecting concepts semantic similarity such as node density, node depth, hierarchical order of nodes and so on, are improved by ontology to accurately define the basic terms and relationships in related fields. Considering the compositive influence factors of concept semantic distance, the relationship between concepts, concept attributes and hierarchical order, an improved ontology-based semantic similarity computation algorithm is put forward. Experiment result shows that the algorithm accuracy is higher than other algorithms.
出处
《桂林理工大学学报》
CAS
北大核心
2012年第2期253-258,共6页
Journal of Guilin University of Technology
基金
国家自然科学基金项目(11061006)
关键词
自然语言
本体
概念距离
概念间关系
语义相似度
natural language
ontology
concept distance
relationship between concepts
semantic similarity
作者简介
作者简介：张兰芳（1964-），女，硕士，副教授，研究方向：教育技术及计算机应用，lfzhang64@163．com
