# Covering Cracks in Content Moderation: Delexicalized Distant Supervision for Illicit Drug Jargon Detection

arXiv: https://arxiv.org/abs/2503.14926 · HTML: https://mlsanity.com/paper/2503.14926/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 21 of its 55 references that are in the map.
- First posted on arXiv 19 Mar 2025.

## Builds on (most cited first, 21 of 21)

- [Efficient Estimation of Word Representations in Vector Space](https://mlsanity.com/paper/1301.3781/) (2013)
- [Don't Stop Pretraining: Adapt Language Models to Domains and Tasks](https://mlsanity.com/paper/2004.10964/) (2020)
- [Distant Supervision for Relation Extraction beyond the Sentence Boundary](https://mlsanity.com/paper/1609.04873/) (2016)
- [Hatemoji: A Test Suite and Adversarially-Generated Dataset for Benchmarking and Detecting Emoji-based Hate](https://mlsanity.com/paper/2108.05921/) (2021)
- [Distantly-Supervised Named Entity Recognition with Noise-Robust Learning and Language Model Augmented Self-Training](https://mlsanity.com/paper/2109.05003/) (2021)
- [Robust Distant Supervision Relation Extraction via Deep Reinforcement Learning](https://mlsanity.com/paper/1805.09927/) (2018)
- [Self-Supervised Euphemism Detection and Identification for Content Moderation](https://mlsanity.com/paper/2103.16808/) (2021)
- [What is Gab? A Bastion of Free Speech or an Alt-Right Echo Chamber?](https://mlsanity.com/paper/1802.05287/) (2018)
- [Detecting the Hate Code on Social Media](https://mlsanity.com/paper/1703.05443/) (2017)
- [Kek, Cucks, and God Emperor Trump: A Measurement Study of 4chan's Politically Incorrect Forum and Its Effects on the Web](https://mlsanity.com/paper/1610.03452/) (2016)
- [Euphemistic Phrase Detection by Masked Language Model](https://mlsanity.com/paper/2109.04666/) (2021)
- [DarkBERT: A Language Model for the Dark Side of the Internet](https://mlsanity.com/paper/2305.08596/) (2023)
- [Searching for PETs: Using Distributional and Sentiment-Based Methods to Find Potentially Euphemistic Terms](https://mlsanity.com/paper/2205.10451/) (2022)
- [VeriDark: A Large-Scale Benchmark for Authorship Verification on the Dark Web](https://mlsanity.com/paper/2207.03477/) (2022)
- [Detection of Illicit Drug Trafficking Events on Instagram: A Deep Multimodal Multilabel Learning Approach](https://mlsanity.com/paper/2108.08920/) (2021)
- [Reliable Decision from Multiple Subtasks through Threshold Optimization: Content Moderation in the Wild](https://mlsanity.com/paper/2208.07522/) (2022)
- [On the Importance of Delexicalization for Fact Verification](https://mlsanity.com/paper/1909.09868/) (2019)
- [Shedding New Light on the Language of the Dark Web](https://mlsanity.com/paper/2204.06885/) (2022)
- [Iterative Delexicalization for Improved Spoken Language Understanding](https://mlsanity.com/paper/1910.07060/) (2019)
- [VendorLink: An NLP approach for Identifying & Linking Vendor Migrants & Potential Aliases on Darknet Markets](https://mlsanity.com/paper/2305.02763/) (2023)
- [Towards Dark Jargon Interpretation in Underground Forums](https://mlsanity.com/paper/2011.03011/) (2020)

## Abstract (arXiv)

In light of rising drug-related concerns and the increasing role of social media, sales and discussions of illicit drugs have become commonplace online. Social media platforms hosting user-generated content must therefore perform content moderation, which is a difficult task due to the vast amount of jargon used in drug discussions. Previous works on drug jargon detection were limited to extracting a list of terms, but these approaches have fundamental problems in practical application. First, they are trivially evaded using word substitutions. Second, they cannot distinguish whether euphemistic terms such as "pot" or "crack" are being used as drugs or in their benign meanings. We argue that drug content moderation should be done using contexts rather than relying on a banlist. However, manually annotated datasets for training such a task are not only expensive but also prone to becoming obsolete. We present JEDIS, a framework for detecting illicit drug jargon terms by analyzing their contexts. JEDIS utilizes a novel approach that combines distant supervision and delexicalization, which allows JEDIS to be trained without human-labeled data while being robust to new terms and euphemisms. Experiments on two manually annotated datasets show JEDIS significantly outperforms state-of-the-art word-based baselines in terms of F1-score and detection coverage in drug jargon detection. We also conduct qualitative analysis that demonstrates JEDIS is robust against pitfalls faced by existing approaches.