【摘要】 提出了一种基于形式概念分析的模式匹配的FCABSM方法,该方法由3部分组成:首先,以朴素贝叶斯文本分类算法为基础设计名称分类算法及描述分类算法,分类目标模式与待匹配模式的元素名以及元素描述,为模式间元素的匹配提供初始依据.其次,利用形式概念分析技术整合分类结果、元素类型信息以及约束信息,提高匹配精度.该阶段为待整合信息创建形式上下文、获取形式上下文中蕴涵的概念、确立概念间偏序关系及构建概念格.最后,以第二阶段的概念格为计算依据,引入基于结构的相似评估模型来计算出最终的匹配结果.实验表明,基于FCA的模式匹配方法的平均性能优于缺少FCA整合的直接匹配方法.更多还原

【Abstract】 A new schema matching approach based on formal concept analysis(FCA) is introduced.The procedure contains three steps.Firstly,the evidence about each element being matched is initialized by applying name classifier and description classifier which are built on Naive Bayes Text Classifier to classify the names and descriptions of the elements.Secondly,FCA is applied to integrate the classified results as well as type messages and constrains to increase the evidence.This step is designed to create formal context for various information to be integrated,acquire the concept contained,figure out the partial order between concepts and construct the concept lattice.At last,a structural similarity measure is introduced to calculate the final matches.Experimental results demonstrate that FCA-based matching outperforms direct matching(without the benefit of FCA).更多还原