Sentiment analysis: Tidy Approach or Qdap's polarity function?


Which approach would you recommend for sentiment analysis?
I'm hesitating between qdap's polarity function or the tidy approach with one of the three dictionaries ("nrc","bing", "afinn").

I like the tidy approach with the dictionaries because it's structured and easier to manipulate, but it doesn't take valance shifters into account like qdap's polarity function does.

Which one would you go for and why?

If you are most interested in an overall tidy data approach to your analysis, the sentiment analysis tools using straightforward dictionaries can be a great place to start and have benefits of being transparent and easy to explore and manipulate. This is especially true if valence shifters aren't a big part of the text you are dealing with (often true in my experience in survey responses, etc).

If you are interested in a sentiment analysis approach that takes into account valence shifters, I recommend @trinker's sentimentr package over qdap. It is DELIGHTFUL to work with :100: and has a great balance of speed compared to the results you get.

1 Like

Thank you very much, for the elaborate response!

Huge, huge fan of your book btw!

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.