Hacker News new | ask | show | jobs
by jtlicardo 1130 days ago
You pretty much summed up the drawbacks of the embeddings approach. In my experience it's pretty hard to extract the relevant parts of text, especially when the text is uniform.