迁移知识辅助的语义稀疏服务聚类方法  

Transferred Knowledge Aided Semantic Sparse Service Clustering

作　　者：田刚[1,2] 何克清[1] 高莹[2] 黄颖[1,3] 

机构地区：[1]武汉大学软件工程国家重点实验室计算机学院,湖北武汉430072 [2]山东科技大学信息学院,山东青岛266590 [3]赣南师范学院数计学院,江西赣州341000

出　　处：《四川大学学报（工程科学版）》2015年第5期116-122,共7页Journal of Sichuan University (Engineering Science Edition)

基　　金：国家重点基础研究发展计划资助项目(2014CB340404);国家自然科学基金资助项目(61202031);江西省自然科学基金资助项目(20142BAB217028);软件工程国家重点实验室开放课题资助项目(SKLSE 2014-10-07)

摘　　要：现有服务聚类方法缺乏对服务描述语义稀疏情境下的研究,因此将迁移学习技术应用到服务聚类领域,尝试解决语义稀疏服务聚类的问题。通过对偶PLSA模型将目标领域和辅助领域语料知识进行融合,利用无监督的方式迁移辅助领域知识,从而提高目标领域语义稀疏服务聚类的能力。实验结果表明,该方法能够提高语义稀疏服务的聚类效果。与K-Means、Agglomerative和PLSA等方法相比,该方法在聚类纯度、熵上均具有更好的性能。The existing clustering approaches are lacking of researching on clustering Web services whose descriptions are semantic sparse. Therefore, a new approach was proposed to makes an attempt to solve the problem of clustering sematic sparse Web service by applying transfer learning method in the domain of Web service clustering. A dual PLSA model was introduced to integrate knowledge of target domain and auxiliary domain which can transfer knowledge using a unsupervised mode to facilitate the process of semantic sparse Web service clustering. Experimental results showed that the proposed method can improve the performanee of semantic sparse Web service clustering. Compared with the approaches of K-means, Agglomerative and PLSA, the proposed approach achieves better perform- ance of the purity and the Entropy.

关 键 词：Web服务聚类 迁移学习 语义稀疏 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]