Update README.md
Browse files# MLSNT: Multi-Lingual Social Network Toxicity Dataset
**MLSNT** is a multi-lingual toxicity classification dataset created via LLM-assisted label transfer, designed to support scalable, resource-efficient moderation in gaming and social media contexts.
This dataset accompanies the paper:
> **Unified Game Moderation: Soft-Prompting and LLM-Assisted Label Transfer for Resource-Efficient Toxicity Detection**
> Accepted at **KDD 2025**, Applied Data Science Track
---
## 🧩 Overview
MLSNT standardizes toxicity annotations across **7 languages** and **15 open-source datasets** using GPT-4o-mini. It enables:
- Binary and fine-grained toxicity classification
- Cross-lingual training with consistent labels
- Production-ready deployment
---
## 🌍 Supported Languages
- **French** (`fr`)
- **German** (`de`)
- **Portuguese (Brazilian)** (`pt`)
- **Russian** (`ru`)
- **Simplified Chinese** (`zh-cn`)
- **Traditional Chinese** (`zh-tw`)
- **Japanese** (`ja`)
---
## 🏗️ Construction Method
1. **Source Datasets**
15 datasets from platforms like Twitter, YouTube, and Reddit, sourced from `hatespeechdata.com` and peer-reviewed publications.
2. **LLM-Assisted Label Transfer**
- GPT-4o-mini used to re-label entries with unified taxonomy
- Only samples where original human and LLM labels agreed were retained
3. **Unified Taxonomy**
- Binary: `toxic`, `non-toxic`
- Fine-grained: e.g., `threat`, `sexual content`, `harassment`, `potentially toxic`
---
## 📊 Dataset Statistics (Post-Processing)
| Language | Total Samples | % Discarded | Toxic % (Processed) |
|----------------------|----------------|-------------|----------------------|
| German (HASOC, etc.) | ~13,800 | 28–69% | 32–56% |
| French (MLMA) | ~3,200 | 20% | 94% |
| Russian | ~14,300 | ~40% | 33–54% |
| Portuguese | ~21,000 | 20–44% | 26–50% |
| Japanese | ~2,000 | 10–25% | 17–45% |
| Chinese (Simplified) | ~34,000 | 29–46% | 48–61% |
| Chinese (Traditional)| ~65,000 | 37% | ~9% |
---
## 💾 Format
Each row in the dataset includes:
- `full_text`: The original utterance or message.
- `start_string_index`: A list of start string indices (start positions of toxic spans).
- `end_string_index`: A list of end string indices (end positions of toxic spans).
- `category_id`: A list of toxic category IDs (integer values).
- `final_label`: A list of toxic category names (string values).
- `min_category_id`: The minimum toxic category ID in the row (used as the primary label).
- `match_id`: A unique identifier composed of the original dataset name and a row-level ID.
---
## 🗂️ Category ID Mapping
| ID | Friendly Name |
|-----|---------------------------------------------|
| 0 | Non Toxic |
| 1 | Threats (Life Threatening) |
| 2 | Minor Endangerment |
| 3 | Threats (Non-Life Threatening) |
| 4 | Hate |
| 5 | Sexual Content / Harassment |
| 6 | Extremism |
| 7 | Insults |
| 8 | Controversial / Potentially Toxic Topic |
## 🔬 Applications
- Training cross-lingual toxicity detectors
- Evaluating LLM-based moderation systems
- Benchmarking multilingual fairness & robustness
---
## 📜 Citation
If you use MLSNT in academic work, please cite:
```bibtex
@inproceedings{yang2025mlsnt,
title={Unified Game Moderation: Soft-Prompting and LLM-Assisted Label Transfer for Resource-Efficient Toxicity Detection},
author={Zachary Yang and Domenico Tullo and Reihaneh Rabbany},
booktitle={Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD)},
year={2025}
}