基于邻接距离属性动态聚类算法采用能综合反映属性名称相似性和语义相似性的“邻接距离”,提高了属性匹配的准确率;以类内损失、类间损失之和最小化为准则,使用动态聚类算法对相似属性进行匹配,不需要设置聚类参数,避免了人为造成的误差。
By means of the "adjoin distance" which reflects name comparability and semantic comparability of attributes, the adjoin distance-based attribute dynamic clustering algorithm improved the veracity of attribute matching. Based on the guide line which minimizes the sum of loss of intra-classes and inter-classes, during the matching of the similar attributes with the dynamic clustering algorithm, it doesnt need to set the parameter of clustering, and avoids the error made by man.