Hello,
I think you are trying to do what this person is doing: Sentiment Analysis on Amazon Reviews - #3 by Andrea123
I’m attaching a modified workflow similar to what I did for Andrea123.
Topic_Modeling.knar.knwf (87.2 KB)
Topic Modeling (Topic Detection) is a fairly difficult task because it requires extensive knowledge of NLP (natural language processing). As well, the data should be related to the topic. From the data you provided, I don’t see price, product, etc often (less than 10 rows for price for example). These comments a bit off topic: “No conoce Bupa”, “Llamar el miércoles a las 5:30, envié correo”, “Cliente es fisioterapeuta de 41 años”. These comments are about products (people?) known, timing, and age, so that is what the model will attempt to extract.
I suggest first watching:
And then watching:
There might be other ways to extract that information (Regex for example).
