I am trying to use Orange 3.20 Text Mining nodes to classify sentiments using a lexicon approach (either using the Method from Liu Hu or Vader).
After selecting the columns, I can see the documents already classified according to positive, negative, neutral or compound (in the example for Vader).
However, I was also trying to extract the most positive and negative n-terms from the documents. An example would be to classify for example the word "staff" that appears in multiple documents according to its average compound score. For example, "staff" could appear as "amazing" in one document (high compound score) and "awful" in another document, and the average in this case could almost be neutral. That would be very important to understand the specific positive/negative drivers in the text.
I'm planning to use Orange for teaching and I do this today with MeaningCloud or Lexalytics and I think it's the only thing that I still cannot do in Orange (without using Python). But probably I'm just new to Orange and not being able to understand how to do it.
Can I do this in Orange without resorting to Python?