NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:7091
Title:REM: From Structural Entropy to Community Structure Deception

This submission deals with privacy issues of revealing community information in graphs and ways to hide this information by modifying the network structure. There is consensus that the methods proposed in the submission are effective for the task of hiding community membership and that the problem is interesting from both a theoretical and practical perspective. For these reasons, I think the paper should be accepted. The reviewers raised some ways in which the paper could be improved. One specific issue is what types of nodes get their attributes hidden. For example, are these "core" nodes in the community with high degree or low-degree periphery nodes? A deeper investigation into how the method gets good performance would be much appreciated.