Coding Open ended questions in R - (Text Classification)

r
text-classification

#1

Hi There,

I am trying to do sentiment analysis for a survey where I have more than 1000 comments. I have few themes like Industry, People, Communication and I need to identify who is commenting on which themes by going through each comment and then tag weather its positive or negative. Is there any way we can customize the list of themes and make our own coding frames for open ended questions. Please help.


#2

Hi @jayanthd

It seems to be an interesting project. It is basically a combination of NLP tasks, i e., text classification (consider themes as classes) and sentiment analysis (positive/negative). So, you can try to solve the problem by doing text classification and sentiment analysis separately.

I am not sure about this part. Could you be more specific?


#3

Thank you so much for the reply. Let me give some more information on the requirement.

Suppose there is feedback from a customer. That feedback should be multilable. It means suppose we receive a feedback on car. The feedback may have response on car design, car price, car segment etc. Depending on the content of the feedback one response should be tagged to multiple category for design, price,segment etc. And each label if response was possitive for design but negative for price. Then each category should have sentiment assigned. In this way we will have multiple label with different sentiment from same response. I am achieving a good accuracy for category but sentiment it’s failing.