aiml-spambot-benchmark

To benchmark spam detection, you evaluate a DistilBERT-based detector against defined social-media campaigns to measure recall. This process applies predefined categories, enforces data integrity, and verifies category coverage using provided datasets.

Spam classification models for moderation are evaluated using recall and per-category results. The benchmark measures how well the detector identifies harmful content across predefined spam categories to gauge robustness and safety.

Yes, the spam-bot benchmark requires the provided spam_tweets.json dataset to evaluate recall. It enforces data integrity by applying fixed thresholds for content length and campaign size during the validation process.

Automated data validation for spam campaign evaluation works by enforcing data integrity, consistent labeling, and checking for placeholder avoidance. It mirrors real-world moderation pipelines by validating category coverage and minimum content length.

Yes, you can use DistilBERT for spam recall evaluation across multiple campaigns. The benchmark workflow validates category coverage and applies fixed thresholds for campaign size to ensure reproducible safety and moderation results.