Agent-based approach for the recommendation and unsupervised classification of enterprise services in the cloud

Abstract

The main objective of this study is to create a collaborative parallel environment, which supports the unsupervised classification and the filtration of an important volume of information. The proposed approach consists in the integration of an agent-based system composed of five reactive agents to assist the recommendation of the stored services in the cloud and to cluster these services through a new improved K-means as well. The conducted experiments and evaluations of the different approaches and measures, such as: Euclidean distance, Manhattan distance and Cosine similarity, show that the proposed approach of the unsupervised classification improve the within cluster sum of squares (WCSS), which facilitates the access to personalized and relevant services requested in a very improved response time, especially through migration to the cloud using agents.