2
$\begingroup$

I am doing a project where I will have to detect emotions from reviews gained by users (through comments). My task is to classify the comments in at least 5 types of emotions that is: happiness, satisfy/unsatisfied, disappointment, sadness, joy. Can I find an already existing model trained for emotion analysis where I can analyze the emotions in each comment. My output needs to give whether the comment is happy or sad/angry etc. I am willing to use the RNN technique. Can someone please help me with my project? Thank you.

$\endgroup$
2
  • 1
    $\begingroup$ The NLP task is called sentiment analysis $\endgroup$ – Erwan Dec 15 '19 at 20:01
  • $\begingroup$ Actually I am not doing sentiment analysis but I need to perform emotion analysis $\endgroup$ – EAnalysis Dec 16 '19 at 5:36
3
$\begingroup$

There are certain pre-trained models available regarding sentiment analysis or emotion analysis like textblob , flair and deepmoji. Here the link to the in the medium -> https://medium.com/@b.terryjack/nlp-pre-trained-sentiment-analysis-1eb52a9d742c

But if you have your own set of emotions for which you want to classify, then the best way would be to gather data with labels (the emotions) and train a model from scratch.

$\endgroup$
3
  • $\begingroup$ Thank you very much Sandy, Do you know any tutorial where it explain step buy steps so that i can train my model from scratch? $\endgroup$ – EAnalysis Dec 18 '19 at 10:20
  • $\begingroup$ If your dataset is such that each text of yours will only have one emotion from the list you mentioned then its a multi-class problem. On the other hand, if you have each text having multiple emotions then its a multi-label classification. Search with these terms you will find appropriate algorithms. For example for multi-label you can follow this - towardsdatascience.com/… $\endgroup$ – Saandeep Sreerambatla Dec 19 '19 at 12:40
  • $\begingroup$ And if you are ok with the answer please accept it. $\endgroup$ – Saandeep Sreerambatla Dec 19 '19 at 15:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.