Research And Implementation On User Modeling Matchmaking In Business Process

HPS(Human Provided Service) is an important research area of SOA(service-oriented architecture). HPS allows people to publish their capabilities and technologies as web services. HPS supports not only the interaction between people, it also supports the interaction between human and software service, which also allows human and software service flexible mixing in bussiness processs.However, due to the difference between human and the uncertainty of services they can provide, there are a lot of difficulties in HPS study, including how to establish an effective user model for human service, develop a flexible and efficient human task allcation mechanism, and achieve human and software services' seamless integration in bussiness process. In this paper, after conducting research and testing on user modeling and human task matching algorithms in HPS, proposed a method for automatic generation of human service, automatic convert user information and registration services to computer readable language; proposed an collaborative filtering algorithm based on user information and interests and an HPSRank algorithm based on interaction context. Finally, on the basis of the aforementioned researches, we implement a conference registration and management system called ConfTool.The main work done by this thesis includes three parts:Firstly, we proposed a mothod for automatic generation of human service.Through research and experments, we abstract user profile related elements for matching human tasks, abstract user research area, education, as well as geographic location and so on, use XML to covert and save user registration information; also we abstract human service, extract the contents of service, service hours and service elements such as input and output, use XML,WSDL,and SCA, convert user registration service to standard web service file-*.WSDL,and save. after testing, the method lets users not aware of specific SOA-related syntax and complex underlying web service running mechanism register service, allows user more convenient and effective to provide service, through element designing, it can also enhance the efficiency of matching algorithm.Secondly, we present an collaborative filtering algorithm based on user information and interests and an HPSRank algorithm based on interaction context. through the introduction of the concept of the sub-word based on statistical, combined with the unique content and features of related knowledge of human service data, proposed an collaborative filtering algorithm based on user information and interests, to facilitate the registration for service; also proposed an algorithm base on PageRank. This algorithm combined with the importance of human service and the relevance to task context, generate the ranking result, to achieve a better human task matching. Test results show the algorithm can basically meet the actual requirements.Thirdly, we designed and implemented a conference management system called ConfTool. The system supports conference organizers to customize meeting processes and manage the actors involved in each process. The system has demostrated a perfect combination of human service and software service, fully embodies the advantages of the concept of HPS.