Visualize sentiment of Hacker News comment threads

There have been more than a couple times that I have found myself reading a Hacker News thread comment by comment, intent on fully understanding the sentiment of each commenter and drawing a conclusion about the post overall

I’ve found digging deep on HN to be useful include when my spouse and I were deciding whether we should take job offers from companies that have been discussed, when doing (very biased) market research on a feature or product, and when learning about new technology

During my most recent comment deep dive, I built a tool to analyze and visualize the sentiment of each comment as a graph and make it pleasant to navigate. I also extracted keyword phrases so they could be counted and compared by sentiment. I used net promoter score (NPS) as a framework for sentiment aggregation

I decided to productize and share my work. HN Sentiment Tool (HST) is a static page built with SvelteKit where you paste a HN URL, paste an OpenRouter API key (you can audit the code or clone and run locally), choose a model, then analyze and visualize the sentiment comment by comment and across the entire post

The tool loads complete with a polarizing thread about Peter Steinberger joining OpenAI to give you an idea of how it works

I hope you find HST a fun and useful research tool. Let me know if you have any questions or suggestions