Vulnerability of Text-Matching in ML/AI Conference Reviewer Assignments to Collusions

Research artifacts of the paper “Vulnerability of Text-Matching in ML/AI Conference Reviewer Assignments to Collusions“, accepted to the USENIX Security 2025 conference. Our artifacts have passed availability, functionality and reproducibility checks as part of the conference's peer-reviewed artifacts evaluation processs. You can download the datasets (dataset.zip), and we also release the **adversarial abstract examples**, code, scripts, and LLM prompts in the reviewer_assignments_vulnerability repo (reviewer_assignments_vulnerability.zip).