This repository contains code for analyzing the problem of detecting reviewer-author collusion rings from bidding datasets. For more explanation on the analyses, please refer to the associated paper: On the Detection of Reviewer-Author Collusion Rings From Paper Bidding.

Before running any code:

• Make empty directories titled datasets/ and results/.
• Download the files from here and place them in the datasets/ directory. The file aamas_2021.csv is sourced from PrefLib. The file wu_tensor_data.pl is sourced from (Wu et al., 2021). The other files are constructed by the scripts construct_authorships.py and synthesize_aamas_text.py, which have additional data dependencies not included in this repository.
• Run the script compile_count_cliques_c.sh to compile the C++ subroutines.

In all scripts, the argument aamas_sub3 refers to the AAMAS dataset and the argument wu refers to the S2ORC dataset from the writeup. Other arguments specify the setting (unipartite/bipartite), size and density parameters, detection method, etc. The following scripts run the analyses:

• clique_eval.py runs the exact clique-counting analyses.
• detection_eval.py runs the detection algorithm analyses. Code for detection methods TellTail and Fraudar was sourced from (Hooi et al., 2020) and (Hooi et al., 2016) respectively.
• success_eval.py runs the colluder success analyses.