README

Code for the paper :

FastCover: A Self-Supervised Learning Framework for Multi-Hop Influence Maximization in Social Networks by Anonymous.

Implementation of an integrated efficient framework solving k-budget constrained d-dominating set problem (k-dDSP).

Link: TBA.

More real-world test graphs (5.88G) are found in Dropbox: https://www.dropbox.com/s/nm2ilieqf8axq2e/data.zip?dl=0.

Important Python Libraries

• igraph=0.9.1
• torch=1.8.1
• dgl=0.6.0 (based on the CUDA version)
• furl=2.0.0
• timeout-decorator=0.5.0

Instructions

File Organization

• data/graphs/: Some small graphs for training and test
• model/: Loss functions and GNN layers (implemented in dgl)
• experiments/: Training and evaluation launchers.
• baselines/: Heuristic algorithms for

Graph Reversed Attention Network (GRAT)

To train a GRAT solving k-dDSP, run the demo experiments/train_models.py.

To evaluate the model, run exeriments/evaluate_models.py