The Community for Technology Leaders
2017 IEEE International Conference on Services Computing (SCC) (2017)
Honolulu, Hawaii, United States
June 25, 2017 to June 30, 2017
ISSN: 2474-2473
ISBN: 978-1-5386-2005-2
pp: 124-131
As the quantity of Web services grow continuously, it becomes more challenging for developers to navigate through and make use of them. Thus, a knowledge map consisting of a summary of individual services and their relations has become increasingly useful. There are two challenges in building such a knowledge graph for Web service ecosystems. First, services keep evolving in terms of function and usage pattern, while their descriptions typically remain static and obsolete. Second, service profiles usually comprise some common background terms which do not differentiate services. To address the two challenges, we developed a novel tailored topic model, named Service Representation-LDA (SR-LDA), to mine effective representations beyond service profiles to build a knowledge map. The key idea is to incorporate mashup descriptions as an indication of service evolution, and introduce a global filter to identify and filter out background terms. Extensive experiments show that the tailored model is more effective than baselines. The methodology of building knowledge maps for the real-world ProgrammableWeb service ecosystem based on the learned representations is also presented, together with the analyses of representative functionality patterns.
Mashups, Ecosystems, Adaptation models, Google, Mathematical model

B. Bai, Y. Fan, W. Tan and J. Zhang, "SR-LDA: Mining Effective Representations for Generating Service Ecosystem Knowledge Maps," 2017 IEEE International Conference on Services Computing (SCC), Honolulu, Hawaii, United States, 2017, pp. 124-131.
98 ms
(Ver 3.3 (11022016))