Community Detection and Matrix Completion With Social and Item Similarity Graphs

In this paper, we design and analyze MC2G (Matrix Completion with 2 Graphs), an algorithm that performs matrix completion in the presence of social item similarity graphs. runs quasilinear time is parameter free. It based on spectral clustering local refinement steps. The expected number sampled entries required for to succeed (i.e., recover clusters graphs complete matrix) matches information-theoretic lower bound up a constant factor wide range parameters. We show via extensive experiments both synthetic real datasets outperforms other state-of-the-art algorithms leverage graph side information.