Web Service Composition System Parameter Library Optimization Algorithm

The widely adopted web service technology poses a new challenge to the composition of services.How to compose related atomic web services to satisfy functionality needs when no atomic web service can satisfy is one of the crucial problems the web service technology needs to solve.Although different composition algorithms are adopted in different composition systems,the matching of parameters is used in most composition systems.Thus,how to improve these match relationship is very important to the usability and capability of composition systems.Based on the analysis of existing composition systems,an optimization method and a group of optimize algorithms are proposed.The main research content includes:This thesis introduces a strategy of parameter matching finding.We analyze the importance of parameter matching in composition algorithms and propose a parameter matching finding strategy which leverages the matching in user's composition requests.User's knowledge of parameter matching is used to improve the matching relationship.This thesis introduces two basic parameter matching finding algorithms:forward analysis and backward analysis.These algorithms are used to find the parameter matching in user's composition requests.First a composition space is constructed based on existing atomic web services and parameter matching relationship.Then the semantic distances between parameters are calculated and leveraged to optimize the multi-service selection.This thesis introduces some mixed analysis algorithm based on two basic analysis algorithms.We analyze the advantages and disadvantages of these algorithms and different adopting scenarios,and how to take advantage of both basic algorithms to improve the efficiency and reliability of the mixed analysis algorithm.This thesis introduces a method to optimize existing parameter repositories in composition systems.The results of analysis algorithms are grouped into equivalence classes and the matching relationships between these equivalence classes are computed.The computing results are filtered according some predefined limits and finally leverage to optimize the parameter repositories and improve the capability of composition systems.