基于特征自适应的本体映射方法

Feature-based adaptive method of ontology mapping

针对现有的多策略映射方法在执行本体映射任务时不能基于本体特征进行动态调整的问题,提出基于特征自适应的本体映射方法(FAMAP).该方法从语言、属性、实例和结构4个维度描述待映射本体的特征,通过计算每项特征相似度的初始度量值和该特征对应映射策略的相似度矩阵得到该特征的置信度,采用Sigmoid方法实现在映射结果组合时基于特征的置信度对该特征对应策略的权值进行自适应调整.在语言级映射策略上,提出基于Wordnet、综合考虑概念信息量和概念抽象度的语义相似度计算方法.实验结果表明,采用FAMAP方法提高了查准率,取得了更好的匹配效果.

Existing multi-strategy mapping methods could not dynamically adjust based on different ontolo- gies and real situation, exposing the inadequacy of exploiting the semantic information and characteristics of ontology. A feature-based adaptive ontology mapping method （FAMAP） was presented to solve these problems. Four dimensions, including language feature, attribute feature, instance feature and structure feature, were adopted to describe the characteristics of ontologies. The reliability of each feature of ontolo- gy was calculated based on the initial metric value of corresponding characteristic and the similarity matrix of corresponding mapping strategy. Then the weight of each mapping strategy can be adapted to the relia- bility of the corresponding feature of ontology. The different mapping strategies were assembled by Sig- mold method, outstanding the part of high prediction. A new semantic similarity method was proposed considering the information content and level of abstraction of concept. Experimental results showed that FAMAP method improved the precision while maintaining the recall of the ontology mapping.