关键词: Web服务生态系统, 消亡服务个体, 复杂网络, 特征提取, 机器学习

Abstract: To improve the benign development of service ecosystem and to forecast potential perishing services individual,the service-tag bipartite graph was established by using complex network,and Service Similarity Network (SSN) based on service tag was built to quantize the competition relationship between services.The Feature Percentage Ranking (FPR) of services was extracted based on SSN properties.A statistic machine learning algorithm was introduced to distinguish perishing services from Web service ecosystem.The application of OpenAPI service ecosystem on ProgrammableWeb showed that the proposed method could predict the potential perishing service individual in system effectively and help users to select services for building up durable service compositions.

Key words: Web service ecosystem, perishing service individual, complex network, feature extraction, machine learning