Show HN: Mood Surf – A topic-based discovery engine for Twitter threads

5 points by researchers 4 years ago · 0 comments · 1 min read

Hey HN! We’ve created Mood Surf, a discovery engine to explore topics that intrigue you. We miss the days you could stumble upon weird, wonderful, human-authored pages on the Web. Twitter threads are similarly niche and personal but so much good content gets lost in the algorithmic timeline.

So we built Mood Surf to explore the conversations that you might not have found otherwise. Click “Shuffle” to jump into a random corner topic space. Click “Dive Deeper” to go down the rabbit hole of similar threads.

Technical details:

We ran graph analysis on the tech/startup-adjacent subset of Twitter graph to find the top contributors within a given niche. We extracted each user’s threads and embedded them using sentence transformers [0]. The “dive deeper” functionality is powered by approximate nearest neighbor lookup using Pinecone [1]

We hope you like the product — share with us any feedback you have!

— The folks at Re:Search

[0] https://www.sbert.net/ [1] https://pinecone.io/

No comments yet.