Decentralized peer-to-peer (P2P) networks can benefit from forming interest-based communities that can provide peers with information about the resources shared in the community and collectively computed rating of their quality as well as about the agents in the community and their reputation. We propose a mechanism for forming communities in a P2P system for sharing academic papers. The mechanism requires each agent to compute its trust in the agents with whom it interacts. A simulation shows that such communities can benefit peers.
Group formation is a diffcult task that arises in many different contexts. It is either done manually or using methods based on
individual users’ criteria. Users may not be willing to ﬁll a proﬁle or
their proﬁle may evolve with time without users updating it. A collaboration may also fail for personal reasons between users with compatible
proﬁles as it may be a success between antagonist users that may start a productive conﬂict inside a team. Existing methods do not take into
account previous successful or unsuccessful collaborations to forge new ones. We introduce a new model of collaborative trust to help
select the “best” ﬁtted group for a task. We also developed one heuristic to ﬁnd the best possible group since in practice considering all
the possibilities is hardly an option.
Existing online mentorship systems typically match mentors and mentees manually. Some of these existing mentorship systems, face-to-face inclusive, do one-to-one pairing and triad mentoring, which involve a senior mentor, a junior mentor and a mentee. However, research has shown that triangulated mentoring, also known as group mentoring, saves resources and performs better than the one-to-one pairing and triad mentoring. This research, therefore, proposes a group online mentorship system which will make use of recommender system, trust and reputation mechanisms. Recommender systems can be used to match-make mentors and mentees and trust and reputation mechanisms can be used to improve the decision process. At the moment, a five-stage process has been proposed and work is still ongoing on the perfect algorithm to use.