An On-Demand Services Discovery Approach Based on Topic Clustering

Abstract

As more and more heterogeneous services are available in the cloud, how to discover services in an efficient and accurate way becomes a key issue. Services clustering is an effective way to facilitate services discovery. On the basis of domain-oriented services classification, this paper proposes a topic-oriented services clustering approach to group the classified services in a specific domain into topic clusters. This can greatly reduce the search space and enhance the efficiency of services discovery. Then the paper proposes a three-phase services discovery approach based on topic-oriented clustering to find more relevant services by ＂domain-topic cluster-service＂ matching. Finally, experimental results on real services demonstrate the feasibility and effectiveness of the proposed approach. The results show that our proposed services discovery approach can efficiently and accurately find users' desired services.