Collaboration Prediction in Heterogeneous Information Networks

To reveal the information hiding in the scholarly big data, relationship analysis among academic entities has been studied from different perspectives in recent years. In this paper, we focus on the problem of collaboration relationship prediction between authors in heterogeneous information networks, and a new method called MACP, i.e., Meta path and author Attribute based Collaboration Prediction model, is proposed to solve this problem. We use a two-phase collaboration probability learning approach. First, topological features with author attributes are extracted from the network, and then a supervised learning algorithm is employed to find the best weight associated with each feature to determine the collaboration relationship. We present the experiments on a real information network, namely the APS network, which shows that our proposed model can generate more accurate results compared with the method only considering structural features.