Research Of Semantic Web Service Discovery Based On Clustering

Currently SOA architecture has been increasingly,in which web services are required.The appearance of a large number of web services brings challenges to the discovery of web services. Traditional web services discovery mechanism based on keywords can not meet the needs of users, how to quickly and accurately find out services in accordance with user requirements has become an important problem to be solved.On the basis of research work of web services discovery methods,a semantic web services discovery method based on clustering is proposed.Firstly,semantic web service technology are reserached. Traditional web services description language does not have the machine understanding semantics, so scholars put forward a web services description language based on semantic,OWL-S which contains Service Profile .Service Profile which provides a basic explanation of service is used in this paper.Secondly,traditional service matching algorithm is analyzed.The semantic similarity algorithms differently based on distance and the amount of information are studied.Considering the merits and demerits of the above two algorithms,a weighted semantic similarity algorithm is proposed.Because this algorithm can take into account the distinction between sons nodes concept and sibling nodes concept in ontology concept tree, so similarity calculation for the whole web services is more reasonable and it can improve the web services in the process of matching performance.Then according to the proposed comprehensive weighted semantic similarity algorithm,web services are measured and analyzed by clustering. In this paper, K-means clustering algorithm is used to get a function similar clustering clusters.Each cluster has a cluster center which represents different functional web services .Finally,on the basis of the above research work,a semantic web service discovery system is developed. The semantic web services matcher is a key part of the system and it mainly uses the algorithm proposed in this paper. According to the ontology in the ontology library, service requester's needs are matched and services whose similarity is more than specified value are selected.Then according to the QoS attributes,the selected services in the priority queue are sorted.Now,the best atomic services are appeared. Experiments show that the method proposed in this paper is better than traditional methods.