In this session, I will review my journey into rebuilding the "Related Stories" feature on Princeton.edu news articles using semantic search technologies. I'll discuss the basics of how semantic search works, how it differs from traditional search, and my iterative approach in implementing a practical solution with the lowest amount of complexity. I'll discuss the iterative approach I took, first from developing a simple proof of concept, researching hosted vector databases, then iterating on it to refine the results and create a production-ready implementation.
This talk is geared towards a more technical audience, but technical content will be kept fairly high level. Everyone can learn something from this talk.
