Research on Service Dependency Mining Technology Based on Tag Extension and Bipartite Graph

Abstract

Due to the rapid development of service portfolio model on network, service computing has become a hot topic in recent years. It is important for both service management and service recommendation to clarify the dependencies between services. Tag, as a widely used network information resource in the current service field, can intuitively reflect the function, purpose and other attributes of the service, and it has become a common resource in the field of service discovery and combination. Tag sparsity, however, will reduce the accuracy of service relationship analysis. Therefore, this paper proposes a service dependency mining method based on tag extension and bipartite graph. By extending the existing tags and abstracting the combination of service and the relationship between the extension tags into a bipartite graph, dependency propagation mechanism in bipartite graph is applied to improve the comprehensiveness and accuracy of the service dependency mining. The programmable Web APIs and Mashups are used to show the validity of the service dependency mining method.