This is a python code for implementing link prediction in two public available cauthorship networks datasets:
1-ArnetMiner
2-dblp
Each cauthorship network is presented as a graph. The graph is generated by running the R code in network_graph_similarity.R script that generate also some similarity features between each two verices/nodes in the network. Those features in addition to other features that are extracted from author network (AMiner_Author_50000) are used to train machine larning classifers that predict links between co-authors in each cauthorship network.
For more detials about this link prediction approach and for uisng this code, please cite the following paper:
Doaa Hassan. "Supervised Link Prediction in Co-authorship Networks Based on Research Performance and Similarity of Research Interests and Affiliations". In Proceedings of the International Conference on Machine Learning and Cybernetics (ICMLC 2019), Kobe, Japan, July, 2019.