Link Prediction Based on Random Forest in Signed Social Networks

摘要

The social network considering both positive (like, support, trust, etc.) and negative (dislike, opposition, distrust, etc.) relationships is called signed social network, the edge sign prediction problem of which has an important application value in the friend recommendation system. We, for the first time, apply random forest to predicting the signs of edges in the signed network drawn from Wikipedia, and achieve the accuracy over 85% (balanced data), which improves the prediction accuracy by nearly 5% over traditional algorithm, indeed, our result has outperformed most existing methods.