Making P2P Downloading Dependable by Exploiting Social Networks

P2P file sharing networks are rife with unusable copies of files, polluted either incidentally or deliberately. Pollution has made P2P downloading undependable - users may end up receiving polluted copies of a target file. To combat pollution, we propose a recommender that maintains users´ opinions on previously downloaded files over their social network and provides recommendations on demand to assist users in making file download decisions. The big challenge for the recommender is to produce right recommendations in face of various potential attacks. To this end, we propose novel techniques in managing user opinions and forming recommendations. Our experimental results show that our recommender greatly outperforms existing download selection strategies in identifying authentic files.