Research And Implementation On Behavior-based Web Service Selection Platform

In recent years, Web Service has been widely applied into many fields such as E-Commerce, E-Government and Electronics-Communication Engineering. Web service plays an important role in these fields. Web Service has also a profound effect on the development and deployment of modern enterprise application. Nowadays, one of the hot spots in Web Service is to select web services that meet the demands of customers among huge web service groups with various structures and functions.Service Orirnted Architecture (SOA) has been adopted to design and implementation of Web Service. Researchers pay more and more attention to do Service Computing in different fields based on SOA. Service Computing, which has been developed in to a technology system, involves service modeling, service description, service discovery, service selection, service composition, and so on. Since service selection is an essential part of service composition, it attracts much attention in current research.There are some methods in Service Selection, such as Trust and Reoutation based methods, OoS(Quality of Service) methods, Function based methods, OWL-S based methods and Service Process based methods, etc. However, few methods take actions and starting resources of service excution process into account in service selection. Then the selected services might kind of fail to meet the needs of clients, and furthermore the quality of provided services might be not so good.To improve the quality of selected services, a method of service selection is proposed in this thesis, which is based on dynamic behavior. Main works of the thesis include:(1) Present a dynamic behavior sequence generation algorithm based on Petri net model;(2) Propose a similarity measure algorithm based on the longest common substring algorithm to define the similarity of two behavior sequences. Based on this similarity measurement, one can choose the web services of higher similarity with client behavior sequence as candidate services;(3) Develop a method of correctness verification method based on Petri net to verify the candidate services. The verified services would be the final selected services;(4) Design and implement a web service selection platform based on the above methods.The proposed Behavior-based Web Service Selection Platform is implemented with MyEclipse environment. It includes 3 function modules: Web Service Formalization module, Dynamic Behavior Sequence Similarity Measure module, and Service Correctness Verification module. System first translates web services into Petri nets by web service formalization module, where the client can edit demand service models and candidate service models. Then web services of higher similarity with client behavior sequences are selected as candidate services sets by dynamic behavior sequence similarity measure module. Services that are presented to users are finally determined through service correctness verification module.