Hacker News new | ask | show | jobs
by nuhbye 3245 days ago
>>We don’t store the files at all, we generate them on the fly at each request.

How do you manage to get right image for a text?

2 comments

Looks like some sort of "sentiment" analysis give the ability to vote/train whatever underlying logic they've built so far.
We have a repository of gifs that are tagged according to our nlp algorithm.