摘要
在信息过载时代,本体映射是本体相似度计算的关键步骤。针对当前本体映射中相似度计算没有充分利用本体富含的语义信息的问题,文中提出一种改进的多策略概念相似度计算方法。该方法从本体概念名称、属性、实例和层次结构四个方面讨论本体概念的相似度,并在计算实例相似度时,提出使用差异度与丰富度两个关键因子解决实例差异问题。最后采用sigmoid函数自动生成各策略结果对应的权重,合并映射结果,提高映射质量。文中采用了两组测试数据与RIMOM方法进行实验对比。实验结果表明,该方法能够有效提高映射结果的查全率和查准率。
Ontology similarity calculation is the pivotal to ontology mapping in the era of information overload. Considering that the current ontology mapping similarity calculation does not take full advantage of semantic information of each issue, an improved multi-strategy concept similarity computing approach was proposed to solve it. The method explores the similarity of ontology concept from four aspects, including ontology concept name, attributes, hierarchies and instances, and in calculating the instance similarity, two key factors, difference degree and richness are used to solve the instance differences problem. Finally, with the sigmoid function, the weight of each strategy are automatically generated and used to combine mapping results to improve mapping quality. Apply two group data to compare with the RIMOM. The exoerimental results show that the method can effectively improve the mapping results recall and orecision.
出处
《计算机技术与发展》
2015年第4期38-42,47,共6页
Computer Technology and Development
基金
陕西省高等学校教学改革研究项目(13BY23)
西安市工业应用技术研发项目(CXY1133(5))
关键词
本体
映射
语义相似性计算
信息增益
ontology
mapping
semantic similarity computing
information divergence
