Service Selection Based on Dynamic QoS Networks
Victor W. Chu, Raymond K. Wong, Fang Chen, Chi‐Hung Chi · 2016
Due to the popularity of web services, similar services yet with different service-level characteristics are commonly available. How to consume these services intelligently has become a challenge. While network-based service selection has been proposed to manage the complexity in recent time, the degree of inter-service connectivity is assumed to be constant for efficiency boost. However, it does vary in the real world. In this paper, we propose dynamic QoS networks to cater for this property in service selection to further enhance service efficiency. Regularization levels are adjusted dynamically according to the variation of the degree of connections with minimal impact to computational performance. Based on historical QoS performances retrieved from service logs, dynamic QoS networks are constructed to reveal the relationships among services. Timely service selections are then guided by recently observed QoS of related services. The effectiveness of our proposed model is shown by experiments.