hanzhaoml / StrategyProof_Conference_Review

Code and data for Strategyproof Conference Peer Review

On Strategyproof Conference Peer Review

This repository contains code and data for paper On Strategyproof Conference Peer Review. A shorter version is present in IJCAI 2019:

Yichong Xu*, Han Zhao*, Xiaofei Shi, Nihar Shah
On Strategyproof Conference Peer Review
Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence, IJCAI-19
​(*: equal contribution)

Please cite the above paper if you use this code or data.

Compare Similarity Scores

1. To compare the similarity score with Divide-and-Rank partition and without:

> cd DivideClusters

> python compare_sim.py

2. To compute the similarity score with random partition: In the same folder run

> python compare_sim.py --compute_random

Use similarity matrix

The matrix is at iclr2018.npz.

import numpy as np
scores = np.load("iclr2018_all.npz", allow_pickle=True)
similarity_matrix = scores["similarity_matrix"]
mask_matrix = scores["mask_matrix"]

Each line in similarity_matrix and mask_matrix corresponds to an author and each column corresponds to a paper. The score between [0,1] represents the similarity, and the mask in {0,1} represents the conflicts.