A Web Service Clustering Method with Semantic Enhancement Based on RGPS and BTM

Abstract

In order to overcome the data sparsity problem in service description text and to improve the web service clustering quality, we propose a web service clustering method with semantic enhancement based on RGPS (Role-Goal-Process-Service) Framework and Bi-term Topic Model (BTM). First, we extend service description text’s feature according to RGPS meta-model framework. Also, we generate the service latent feature by BTM. Then, we employ K-means on the generated features. The results of experiments on service registry PWeb show that this method can get better clustering performance in purity and entropy. It is proved that this method has great efficiency compared with the baseline methods K-means, Agglomerative and LDA (Latent Dirichlet Allocation). This paper enhances the service clustering performance and creates foundation work for service organization and recommendation.