Fast teaming on P2P networks based on expertise and relation

In order to explore the ideas of combining expertise and relation in team formation via P2P networks, this study developed a communication platform based on P2P networks for running a limited size of a virtual community. There are three objectives of this study. First, this study attempts to overcome such limitations as arm length relations, personal attributes, and vulnerability of client-server systems. Second, the team formation method is designed to improve the efficiency in the forming process as well as the fitness of team members. Third, this study demonstrates the strength of distributed network system in quickly allocating team members. This study conducted two comparative methods to evaluate the strength and weakness of the proposed method. The control group is called "FART method" and the experimental group is called "SFART method". FART method uses fuzzy ART to learn to assign peers to certain subtasks based on the demand and possessed expertise for subtasks. The input attributes to fuzzy ART network only consider peer expertise. SFART method uses fuzzy ART network to recommend candidate peers to execute subtasks by considering both expertise and relation tie between participant peers. The experimental results show that the proposed mechanism has successfully reduced the traffic cost during the team formation process while maintaining the fitness of the team members.