TORA: Topic-based Rule Affinity for Content Moderation

This repository provides datasets and LLM prompt templates used in the paper:

TORA: Topic-based Rule Affinity for Content Moderation in Social Media Communities

---

📌 Overview

This repository supports research on rule-aware content moderation by releasing:

• 📊 A curated dataset of Reddit posts and community rules
• 🧠 Prompt templates for Large Language Models (LLMs) used in:
  • Topic generation from community rules
  • Content moderation baselines (zero-shot / few-shot)

⚠️ Note: This repo does NOT include trained models, only data and prompt templates.

---

📂 Repository Structure

├── data/
│   ├── posts.csv
│   ├── rules.csv
│   ├── labels.csv
│   └── README_data.md
│
├── prompts/
│   ├── topic_generation_prompt.txt
│   ├── moderation_zero_shot.txt
│   ├── moderation_few_shot.txt
│   └── README_prompts.md
│
├── examples/
│   └── sample_input_output.json
│
└── README.md