|
I have collected a number of Amazon ratings and comments. Each comment is about a product and is accompanied by a numerical rating (from 1-5). I am wondering if there are articles about inferring the numerical rating from the text? Also I am wondering what would be the best way to properly featurize the text (bag-of-words, bigrams, POS, ...) Thank you |
|
This sounds like a sentiment analysis problem, which can probably be extended to a regression analysis without too much trouble (in the same way that classification can be in some circumstances). This paper seems to look at your problem. |
