A Novel Outlier-Tolerable and Predictive Approach to Web Service Composition

Abstract

The Quality-of-Service (QoS) aspects of Web service has gained popularity in the field of service computing. QoS-oriented Web service composition is a distributed model to construct new web service on top of existing primitive or other composite web services with QoS guarantees. A major challenge in this field is that the QoS data of candidate services are with run-time fluctuations and thus difficult to predict. Traditional approaches in this direction tended to address the challenge by statistics, prediction and neural network-based models. A major limitation of these methods lies in that they ignore outliers data in the historical QoS data, in terms of inconsistencies, errors, shifts, corruptions, etc. In this work, instead, we consider outliers in QoS series to be non-neglectable, and propose an outlier-tolerable and predictive approach to service composition through leveraging a joint estimation-based outlier detection method and a niched genetic algorithm. To validate the effectiveness of our proposed method, we conduct extensive case studies based on different outlier conditions, and the experimental results show that our method is superior to existing ones.