aiml-spambot-benchmark

What problem does it solve?

This Skill provides a structured benchmark workflow to evaluate how well a DistilBERT-based spam detector detects harmful content in social media data, aligning model evaluation with safety and moderation goals.

Core Features & Use Cases

• End-to-end spam-bot benchmark that validates category coverage, placeholder avoidance, and minimum content length across multiple campaigns.
• Automated checks mirror real-world moderation pipelines, enforcing data integrity, consistent labeling, and presence of coordination signals.
• Reproducible evaluation of recall across predefined spam categories to gauge robustness and safety.

Quick Start

Run the benchmark with the provided spam_tweets.json to evaluate recall and ensure category coverage.