Web Service Recommendation And Discovery Based On Relationships Between Services

Service Oriented Computing (SOC) is one of the most popular topics in current software field. SOC advocates supporting the openness of software system in a standard way. It provides the services and collaborative management which makes the complex business system doing well with its new products, and improves the software development efficiency. Service Oriented Architecture (SOA) enables the distributed applications having better flexibility and reuse ability. While Web service as the mainstream of the current SOA implementations, the number is growing at a super linear rate. Facing with the rapid expansion of the Web service resource, how to find the services from the large service resources conveniently and accurately has become a great challenge in industry and academia.Fortunately, the technology of Web service recommendation and service discovery have show us the direction to solve the current diffcult problem of finding services.Current popular recommendation technology is widely used in the field of electronic commerce, but once applied to the field of Web service recommendation it returns out a bad result. It’s because of the heterogeneity of Web service and the diversity of user requirements. Therefore, how to carry out the service recommendation is one of the difficulties in the field of Web service. While making the Web service recommendation most existing technology always start from the similarity between services and users. But, this method has mostly ignored the inner association and the compatibility between services which makes the recommended Web service may not be able to cooperate well with users’ services. It makes a waste of the web service resources and contrary to the original intention of SOC software reuse. The recommendation technology of Web services can be used in the case of not understanding users’ requirement, while Web service discovery is used in the case of understanding users’ requirement. Because of the Web service description documents lacking of semantic information, service discovery’s precision and recall is low. Web service discovery based on ontology research is not mature, while researchs on the association rules, and the text clustering method based on vector space model(VSM) achieves great success. Based on the correlation between Web services, this pape starts from Web service clusting and VSM, and points out the deficiencies of current researchs.Lastly, this paper put forward new solutions.The main research content and innovation are as follows:(1) Firstly, we put forward a method of Web service recommendation based on association rules mining(RecARM), and improve it’s algorithm. In this paper we make use of the historical combination information of Web services. And construct the association rules.Then combined with users’ existing Web services to make recommendation and help to improve an optimzed the service compositions for the users. At last, we made a experiment which result shows a better stability and accuracy of this recommendation method.(2) Secondly, we proposed a Web service discovery method based on clustering. To solve the existing problems of common clustering method,we make a improvement of the algorithm. And finally solves the problem of uncertainty of the K value(The initial clustering number).This paper provides new methods an ideas for the Web service recommendation and discovery.