Abstract

Inspired from ​1 which efficiently discovers groups in peer to peer systems. This paper talks about implicit groups based on the person's stated attribute, either discovered through download history, or stated otherwise. From research ​1 and ​2, community directed searches have proven to be quite efficient. We would like to work on explicit groups which are searchable as well. This will not only bolster content sharing among already existent members, but would also attract new users who will be enticed into participation on seeing an active and successful group. Also, we intend to exploit social connections of a peer, by either inviting, or sharing group activities. This is because in social networks, one of the main motivations of joining groups, or getting involved in any activity, comes from one's (being peer here) family and friends. As the group evolves from a single person to multitudes of people, it ends up with all types of people. Some are active, some passive; some are beneficial, some are malicious. In order to protect the group from spamming and malicious content, moderators can be selected. These moderators can be trusted based on their earlier behaviour, and their general outlook. Selection of such moderators can be done by using an experience function ​3 of a peer in the context of group activities.