Research And Implementation Of Web Service Discovery Based On Process Model

In recent years, with the development of web services, how to find efficient and accurate web services that meet the user demands from amounts of candidate services has become one of the urgent problems to solve in Web Computing. The grammar level methods of service discovery are usually based on keyword matching, which might lead to low accuracy of discovered services. The semantic level methods are usually based on functional matching. Both methods do not take service process matching into account, so that they might have low accuracy, recall rate or precision.Aiming at the problems above, the thesis studies process-based web service discovery methods, in which functional matching and service process matching are both considered. Main works include:(1) Propose a service model converting method to convert the service into process model tree.(2) Develop a similarity matching algorithm based on process model service, including the 10 matching algorithm of atomic service and structure matching of service algorithm.(2.1) For the similarity matching algorithm of atomic service, first, combining logical reasoning and ontology tree semantic distance, a concept similarity measure method is proposed to match the single input and output parameter of atomic services. Then, choose the optimal matching parameter set for further computing.(2.2) For the similarity matching algorithm of service structure, the algorithm first do the two process model trees isomorphism judgment. Then the similarity of service names is calculated using a string similarity algorithm GST.(2.3) Integrate the atomic service similarity and the structural similarity to calculate the similarity between request service and provided service.(3) Based on the above methods, a process model based service discovery system is designed and implemented. The system includes atomic services matching module and service structure matching module. The system is implemented under Eclipse environment, Jean reasoning machine, and Wordnet domain ontology. The experimental results show that the proposed algorithm can improve the accuracy of web service discovery.