Asymmetry of social interactions and its role in link predictability: the case of coauthorship networks

This archive contains datasets (DBLP, Cond-Mat, APS) used in the article "Understanding weight-topology correlations in social networks: The case of coauthorship networks" by Kamil Orzechowski, Maciej Mrowinski, Agata Fronczak and Piotr Fronczak. All files have the same format. Fields are separated by space: [FirstNodeID] [SecondNodeID] [weight(symmetric or asymmetric)] The archive also includes a Jupyter notebook with Python code that enables the computation of symmetric and asymmetric overlaps as well as link prediction using the four predictors discussed in this publication. At present, the notebook is adapted to analyze data from the Cond-Mat database. To use it for the APS or DBLP database, change the file names in the first cell in line 11 and 19. Take into account that for the last two datasets, the computation time may be extended from several seconds to several hours.