An On-demand Service Discovery Approach Based on Mined Domain Knowledge

With the increasing availability of Web services, how to effectively and accurately discover services according to users´ requirements becomes a key issue. In this paper, we propose an on-demand service discovery approach based on “requirement-domain-topic cluster” matching. The proposed approach is achieved by the following steps: domain-oriented service classification based on an ontology-empowered Support Vector Machine (SVM), topic-oriented service clustering based on Latent Dirichlet Allocation (LDA), and on-demand service discovery based on the mined domain knowledge. The proposed approach will contribute to the management of domain services, which can greatly facilitate on-demand service discovery.