Using user similarity to infer trust values in social networks regardless of direct ratings | IEEE Conference Publication | IEEE Xplore