This repository contains the implementation for the RedThread algorithm from this paper.
There are 2 non-trafficking related datasets used in the experiments Discogs and Memetracker which can be downloaded from their respective websites. The trafficking data can be shared separately via a google drive link.
Once you have the data file, to first obtain the data features in the required format, run the code in the exploratory_analysis.ipynb file. This code will create the sample data files inside data folder.
Following this, run redthread_run.py to run the RedThread algorithm. The default data path is set to data/sample_data/ which points to the files created in Step 2. In order to use another path, provide the path as a command-line argument. To see other arguments, run python redthread_run.py --help. The current code runs very slowly and needs to be improved for efficiency. Due to the large size of the data, they have been uploaded here
The graph models built after Step 3 will be stored in the models folder.