Aspect based sentiment analysis libraries

Question

I'm working on a project where I have to perform aspect-based sentiment analysis on verbal comments. Can anybody suggest some good existing libraries or examples?

Answer 1

Something similar to your project is the Twitter sentiment analysis projects.

Follow some examples:

1. First example
2. Second example
3. Third example

I suggest you to use NLTK library. Check also the 'How To Section' for examples.

I hope that these information are usefull.

If these are not in your interest, please add some details to your questio in order to answer better.

Answer 2

Not a tool per se but I had a a similar project and got pretty good results using the methods outlined in this paper. The basic idea is to use something like CoreNLP to run a dependency parse and then use some predefined patterns(like NN is/are/was ADJ) to find (aspect, adjective) pairs. The adjectives are then assigned a sentiment based on a provided adjective sentiment lexicon.

I was working with Yelp restaurant reviews and was able to code up a reasonably accurate extractor within a few days.