I'm running Google's NL API sentiment analysis on Twitter content, and am not clear on how to focus the sentiment analysis in situations where the tweeter is commenting on a post or article they're sharing or retweeting.
I'm currently running sentiment analysis on the main tweeter's status text. However, sometimes the entity for the sentiment analysis is only mentioned in the referenced tweet, which then throws off the sentiment analysis when it can't find the entity.
Ideally, I'd like the sentiment analysis to assess both the main tweet and the referenced tweet, and understand how the main tweeter's attitude might be relative to the referenced tweet.
So if the referenced tweet is negative, but the main tweet responds negatively to that negativity, it could actually be positive sentiment towards the entity.
Any input on the feasibility of that would be greatly appreciated.