基于相似度的关系模式和本体间映射算法

A Mapping Algorithm between Relational Schemas and Ontologies Based on Similarity Calculation

为了解决关系数据库模式和本体间的集成问题,存在映射关系,针对模式映射算法,提出了一种基于概念间相似度的计算策略,在相邻结点的相似度推断当前结点的相似度的过程中,根据结点相邻元素的数目对结构化相似度的计算赋予不同的权重;同时为了提高算法的收敛速度,在迭代过程中将初始值作为迭代值的一部分加入迭代过程中的相似度,并使用两次迭代结果的线性相关系数作为判断收敛的依据。实验结果表明,改进后的算法在映射的精度和收敛速度上都优于同类型的其它算法。

To solve the data integration problem between relational databases and ontologies, the paper proposes an improved mapping algorithm which takes a relational schema and ontology as input and uses iteration for measuring similarity. In this algorithm, different weights are attached for mapping elements according to different number of adjacent elements. Furthermore, the paper adds initial similarity value as a part of structural similarity value in the iter ation, and uses linear correlation coefficient as a criterion of termination for convergence acceleration. Experiments show that this approach outperforms other algorithms of the same category in accuracy and efficiency.