I analyzed 571M Amazon reviews to find the most profanity-filled customer rants
burla-cloud.github.io
April 27, 2026
2 min read
🔥🔥🔥🔥🔥
44/100
Summary
A Burla cluster processed 275 GB of Amazon reviews using 1,000 CPUs, categorizing them by various expressions of extreme sentiment. The system can display raw content filtered for profanity, rants, and other intense reactions.
Key Takeaways
- A Burla cluster processed 275 GB of Amazon reviews across 34 categories, identifying unhinged content using various profanity and rage metrics.
- The analysis produced a categorized dataset that includes profanity, slurs, and extreme emotional expressions in reviews, with results available for public access on GitHub.
- The methodology involved multiple processing passes to refine the identification of strong profanity and emotional content, resulting in a final "Unhinged corpus."
- Users can reproduce the analysis on their own Burla cluster in approximately 15 minutes using provided scripts.
Community Sentiment
Positives
- The creativity in the reviews showcases a unique form of expression, highlighting the diverse ways customers articulate their experiences.
- The lack of censorship in the comment section contributes to a humorous and engaging atmosphere, enhancing user interaction.
Concerns
- Concerns arise about the implications of using profanity-laden reviews as training data for AI, potentially influencing model behavior in undesirable ways.