Prediction of missing links is an important task in many application domains such as google ranks, wikipedia links etc. Many methods have been developed for this task and their performance can be influenced by the structure of networks. However, little work has been done to compare the performance of these methods for networks that are heterophilic or homophilic and contain unbalanced groups size.
In this project, the student uses the generative model of networks that contain homophily and group sizes , and evaluates the performance of link prediction methods for these networks (similar to ). The results later will be compared with real network data such as wikipedia.