Do ya’ll have any thoughts on the best approach for large scale sentiment analysis for social media posts? I don’t know much about it, so any advice or direction is appreciated. Thanks!
Do you have a corpus of data and some categories? Are you just looking for positive/negative?
I just watched that video, very helpful!
I want to use the ‘polarity’ of positive and negative. IE, a scale from 0-1, or 1-10. So that is one consideration.
Another consideration is how to scale it to millions of comments while minimizing compute resources.
I like the method of random sampling the data set a few times, then combining the results. Could probably get a p value for verification sake.
Yeah so basically you just keep taking random samples until you get convergence. That’s one way