Analysis on large-scale rating systems based on the signed network

In many user-object online rating systems, it is obviously that the ratings usually show the users' attitudes: like or dislike the objects. Inevitably there is a need to introduce the sign into the rating systems. We first focus on how to construct signed bipartite networks on rating systems and reveal the basic properties of them. We also analyze the basic motif of signed bipartite networks: quadrangle. Then we introduce a novel projection method Signed Common Neighbors (SCN) to get the projection to signed user-network. The basic statistics of the projections show that SCN can well reflect the roles of negative edges.