一种基于特征信息组合的本体匹配流程  

An Ontology Matching Process Based on Combination of Feature Information

作　　者：邹黎君[1] 凌兴宏[1] 姚望舒[1] 

机构地区：[1]苏州大学计算机科学与技术学院,江苏苏州215006

出　　处：《计算机工程》2013年第9期271-276,共6页Computer Engineering

基　　金：国家自然科学基金资助项目"基于逻辑强化学习的Deep Web模式匹配研究"(61070122);"基于tableau模型的逻辑强化学习理论及方法研究"(61270223)

摘　　要：在串行和并行的多策略本体匹配流程中,存在重复计算、匹配效率不高和多个相似度矩阵难以聚合等问题。为此,提出一种基于本体特征信息组合的混合多策略本体匹配流程。利用本体的名称、注释、实例等元素级信息进行初始匹配,并直接确定一部分实体的匹配关系,对于无法确定匹配关系的实体,则再利用本体的内外部结构特征迭代计算其相似度进行判断。实验结果证明,该方法能有效减少匹配的计算量,同时特定信息的组合能保证匹配准确度,具有较好的匹配效果和效率。In the serial and parallel multi-strategy ontology matching process, there exits the problems of repeated calculation, low matching efficiency and difficulty of polymerizing multiple similarity matrixes. This paper proposes a new multi-strategy ontology matching process based on the combination of ontology feature information. Making use of the element level information such as name, comment, instance to do the initial matching, during this stage part of the entities＇ relation can be determined. For the ones that can＇t be determined to be matched, their internal and external structure information can be used to calculate the similarity between them. Experimental results show that this method can reduce the matching calculation effectively, and at the same time, the combination of specific information can ensure matching accuracy. It also proves the method has good matching effect and efficiency.

关 键 词：语义WEB 本体 多策略 本体匹配 特征信息组合 匹配流程 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]