摘要
在目前通用的Web服务描述标准WSDL基础上,文中提出一种改进的操作相似性度量方法MOSM.MOSM在数据预处理后将Web服务内含的操作(operation)建模为无序标签树,并通过计算满足约束的编辑距离对其进行相似性度量.其具体做法是抽取操作的XML模式的树形结构,对结构进行变形,只保留标签结点;然后计算生成的无序标签树之间满足约束的编辑距离,将操作相似性度量的问题转化为无序标签树匹配的问题.该文的创新主要在:建模时采用满足约束的无序树模型,在编辑距离算法中引入支持不对称性的代价模型,另外为结构和文字标签匹配引入了相似系数.文中最后给出实验了对比结果,MOSM算法能有效提高top-k查准率,对找寻相似的备选操作具有重要意义.
Based on the current Web service description standard WSDL （Web Service Description Language）, a Modified Operation Similarity Measure （MOSM） method is proposed. After the data preprocessing, an operation included in a Web service is modeled as an unordered labeled tree and get the similarities measured with other operations by calculating the constrained edit distances in MOSM. First the tree structure of the operation＇s XML schema is extracted and transformed so as only the tag nodes are kept; then the constrained edit distances among the generated unordered labeled trees are calculated, which turns the operation similarity measure problem into the unordered labeled tree matching problem. The innovations of this paper mainly are. Using the constrained unordered trees instead of the ordered trees as the model, introducing the asymmetric characteristic into the cost model of the tree edit distance algorithm, and introducing the similarity weight into both the structure matching and the string label matching. In the end of the paper the experiments shows that MOSM can effectively improve the top-k precision, which makes much sense to look for the similar back up operations.
出处
《计算机学报》
EI
CSCD
北大核心
2008年第8期1331-1339,共9页
Chinese Journal of Computers
基金
国家科技基础条件平台“大型科学仪器设备资源的建设与整合”(2005DKA10100)资助~~
