论文部分内容阅读
如何从海量的Mashup服务集中快速、准确的找到满足用户需求的Mashup服务,成为一个具有挑战性的问题.在M ashup服务发现中,预先对M ashup服务进行聚类,将大大缩小服务搜索的空间与范围,提高M ashup服务发现的效率与精度.本文提出一种新颖的融合K-Means与Agnes的Mashup服务聚类方法(MSCA).该方法,首先对Mashup服务中的Tag标签进行扩充和排序;其次,计算Mashup服务的集成相似性;接着,应用K-Means算法对Mashup服务相似度矩阵进行聚类,找到相似度较高的Mashup服务将其划分到N个原子簇中,再利用Agnes算法对N个原子簇进行层次聚类.最后,从Programmable Web上爬取了13082个Mashup服务作为实验对象,实验结果表明:相比传统的基于K-Means算法的Mashup服务聚类方法,MSCA方法的平均查准率和查全率分别提高了5.18%、5.84%,切实提高了服务聚类及发现的精度.
How to quickly and accurately find a Mashup service that meets the needs of users from a large number of Mashup services is a challenging problem.Classifying the Mashup service in Mashup service discovery in advance will greatly reduce the space for service search And improve the efficiency and accuracy of Mashup service discovery.This paper proposes a novel Mashup service clustering method (MSCA) that combines K-Means and Agnes.This method first expands and sorts Tag tags in Mashup service Secondly, we calculate the similarity of Mashup service. Then, we use K-Means algorithm to cluster Mashup service similarity matrix and find Mashup service with high similarity and divide it into N clusters. Then we use Agnes algorithm Clustering N atomic clusters.Finally, 13082 Mashup services were crawled from the Programmable Web as experiment objects, the experimental results show that compared with the traditional clustering method based on K-Means algorithm for Mashup service, the MSCA method The average precision and recall rate increased by 5.18% and 5.84% respectively, effectively improving service clustering and finding accuracy.