Using Support Vector Machines for Intelligent Service Agents Decision Making.

Analyzing online web services' behavior is a difficult task. The challenge is to effectively manage their cooperation and group work while as rational systems they always seek to maximize their overall utilities. Existing approaches either manage to enhance the quality of service provided by web services or group them together to corporate a stronger web service by gathering web services of the similar functionalities. Although these approaches have shown to be useful, they are not practical in the sense that enhancing the quality of service is a costly process that is not always the best option. In this paper, we present an efficient approach that applies support vector machine to equip web services with this machine learning algorithm to train the previously generated data and effectively make decisions to cooperate with one another. In our experiments, we applied three kernel functions to create the normal model and compared their overall performance together as well as the benchmark, that is, rational web services without learning abilities. The results show that the Gaussian kernel outperforms the other two learning models as well as the benchmark non-learning model by maintaining high true join recommendations rate while producing low false not-join recommendations.