GraphSAGE Implementation for Characterizing and Detecting Hateful Users on Twitter

Requirements

pytorch >0.2 is required.

Running examples

1. Download the dataset from kaggle.

2. You need to create folders hate and suspended as follows:

   |- hate
   |---- users.edges
   |---- users_all_neighborhood.csv
   |---- users_hate_all.content
   |---- users_hate_glove.content
   |- suspended
   |---- users.edges
   |---- users_all_neighborhood.csv
   |---- users_suspended_all.content
   |---- users_suspended_glove.content
   

3. Execute python -m graphsage.model to replicate everything.