摘要
现有的基于进化算法的本体映射技术在面对大规模本体映射问题时,由于搜索空间太大导致算法效率低下,从而使其无法有效地在实际中得到应用。针对这一问题,提出了基于快速非支配排序的多目标遗传算法(NSGA-Ⅱ)的大规模本体映射方法。该方法通过三个步骤来映射本体:1)通过基于邻居相似度的划分算法来将源本体划分为不相交的概念块;2)通过相关概念过滤方法来确定目标本体中同源本体概念块相关的概念块;3)使用NSGA-Ⅱ方法来完成概念块之间的映射并通过贪心算法集成最终的结果。使用OAEI 2012的小规模的书目本体测试数据集和大规模的生物医学本体测试数据集对所提出的方法进行测试。同OAEI 2012的参与者的比较结果表明,所基于NSGA-Ⅱ的大规模本体映射方法能够在较短的时间内获取较好的本体映射结果,因此该方法是有效的。
The application of existing ontology aligning technologies based on evolutionary algorithm is limited by the huge search space of large scale ontology aligning problem. To this end, in this paper, a large scale ontology aligning approach based on a fast elitist Non-dominated Sorting Genetic Algorithm for multi-objective optimization （ NSGA-Ⅱ ） was proposed. To be specific, it worked in three steps： 1） a neighbor similarity based ontology partitioning algorithm was presented to split the source ontology into a set of disjoint concept blocks; 2） a relevant concept filtering method was proposed to determine the concept block in target ontology associated with each source one; 3） NSGA-Ⅱ was utilized to align the various concept block pairs and a greedy algorithm was used to aggregate various results. Small scale bibliographic ontology benchmark and large scale biomedic ontology benchmark in OAEI 2012 were used to test the proposed approach. The comparisons with the participants of OAEI 2012 show that the large scale ontology aligning approach based on NSGA-Ⅱ is able to determine good alignments in a short time, and therefore it is effective.
出处
《计算机应用》
CSCD
北大核心
2014年第6期1622-1625,1630,共4页
journal of Computer Applications
基金
福建省教育厅科研项目(JA13227)
关键词
大规模本体映射
本体划分算法
快速非支配排序的多目标遗传算法
large scale ontology aligning
ontology partitioning algorithm
fast elitist Non-dominated Sorting Genetic Algorithm for multi-objective optimization （NSGA-Ⅱ ）
