Reputation: 61
I am working on Natural Language Processing and having a hard time on analyzing sentiment values on certain entities. For example if i say: "I love the fact that I am finally leaving Chicago", it says that the sentence is a positive sentence. But as you can see it is pretty negative for the entity "Chicago". Anyone knows if there is any related work on this? Thanks in advance
Upvotes: 1
Views: 293
Reputation: 61
The key word for this is “targeted sentiment analysis”
There are many articles on it and also code on GitHub.
I’ll update if I find the link for the model I used.
You might need to do some labeling and training for your specific requirements.
Update: https://github.com/songyouwei/ABSA-PyTorch
train data will look like this:
I love the fact that I am finally leaving $T$, Chicago, -1
Upvotes: 3