MTTM: Metamorphic Testing for Textual Content Moderation Software

TL;DR

This paper introduces MTTM, a metamorphic testing framework for evaluating and improving textual content moderation software's ability to detect toxic content, especially when malicious users make minimal modifications.

Contribution

The paper presents a novel metamorphic testing approach for textual moderation software, including relations and perturbations, and demonstrates its effectiveness in revealing vulnerabilities and enhancing model robustness.

Findings

MTTM detects up to 83.9% errors in commercial moderation tools.

It achieves up to 91.2% error detection on state-of-the-art algorithms.

Retraining with MTTM-generated cases improves model robustness significantly.

Abstract

The exponential growth of social media platforms such as Twitter and Facebook has revolutionized textual communication and textual content publication in human society. However, they have been increasingly exploited to propagate toxic content, such as hate speech, malicious advertisement, and pornography, which can lead to highly negative impacts (e.g., harmful effects on teen mental health). Researchers and practitioners have been enthusiastically developing and extensively deploying textual content moderation software to address this problem. However, we find that malicious users can evade moderation by changing only a few words in the toxic content. Moreover, modern content moderation software performance against malicious inputs remains underexplored. To this end, we propose MTTM, a Metamorphic Testing framework for Textual content Moderation software. Specifically, we conduct a pilot study on 2,000 text messages collected from real users and summarize eleven metamorphic relations across three perturbation levels: character, word, and sentence. MTTM employs these metamorphic relations on toxic textual contents to generate test cases, which are still toxic yet likely to evade moderation. In our evaluation, we employ MTTM to test three commercial textual content moderation software and two state-of-the-art moderation algorithms against three kinds of toxic content. The results show that MTTM achieves up to 83.9%, 51%, and 82.5% error finding rates (EFR) when testing commercial moderation software provided by Google, Baidu, and Huawei, respectively, and it obtains up to 91.2% EFR when testing the state-of-the-art algorithms from the academy. In addition, we leverage the test cases generated by MTTM to retrain the model we explored, which largely improves model robustness (0% to 5.9% EFR) while maintaining the accuracy on the original test set.