无约束相似性.doc

结合属性分布特征的模式匹配算法* *本研究得到863高技术研究发展计划的项目资助（项目编号：2007AA01Z438)。 王宇1,2, 方滨兴1, 吴博1,2,宋林海1,2,郭岩1 1中国科学院计算技术研究所智能信息与智能安全中心, 北京, 100190 2中国科学院研究生院, 北京, 100190 E-mail: wangyu2005@ 摘要：本文提出了一种结合属性分布特征的Web模式匹配算法，属性分布特征包括属性对互斥特征和属性对共现特征。属性对互斥特征由属性对的互斥性和出现次数计算得出，这个特征隐含了属性对的语义相似程度。为了充分利用传统的属性名、属性值相似性特征，本文通过机器学习方法结合属性对互斥特征与相似性特征进行属性匹配。并以潜在的匹配属性对为基础，引入有约束的属性聚类方法进行Web模式匹配，聚类方法的约束条件来自属性对共现特征。实验结果表明，相对于仅使用相似性特征的方法，结合属性分布特征的Web模式匹配算法取得了更好的结果，解决了单独使用属性名相似性能处理的属性较少，而属性值相似性需要针对特定领域优化的问题。 关键词：属性对互斥，属性对共现，Web模式匹配，约束聚类 Schema Matching Incorporating With Attribute Distribution Features Yu Wang1,2, Binxing Fang1, Bo Wu1,2,Linhai Song1,2 and Yan Guo1 1 Institute of Computing Technology, Chinese Academy of Sciences, Bei 2 Graduate School of Chinese Academy of Sciences, Beijing 100190 E-mail: wangyu2005@ Abstract: This paper presents a new Web schema matching algorithm which incorporate attribute distribution features. Attribute distribution features include the mutually exclusive feature and the co-occurring feature. By discovering mutually exclusive attribute pair and various statistics of the attribute pair, the mutually exclusive feature is calculated which implies the semantic similarity of the attribute pair. To utilize name similarity and value similarity based features, the attribute distribution features are combined with traditional similarity based features through machine learning techniques. After potential matched attribute pairs are discovered, this paper introduces the co-occurring feature as the constraint of clustering algorithms and solves the web schema matching problem by constrained attribute clustering algorithms. Experiments on a wide variety of domains demonstrate our method performs better than methods using only similarity based features: find more potential matched attribute pairs than using only name similarity, and does not over-optimized towards any predefined domain while value similarity based features does. Keywords: Mutually Exclusive Attributes, Co-occurrin