Sentiment Analysis

SilkeKoch

Hi everyone,  
I am quite new to rapidminer and text analysis and I wonder how to find out which words are important for the positiv or the negativ class. Do I really just look at the word vectors? I learned in an rapidminer tutorial that you can use Naive bayes and you get the distribution table. 
I do use the sentiment template from rapidminer which uses SVM and replaced it with naive bayes but did not get the distribution table.

MartinLiebig

Hi,

if it is a linear svm, then it has a "wei" port which can be used Further the operator "Explain Predictions" may be very helpful for you.

Best,

Martin

Telcontar120

You could also separately run a Weight by (choose your favorite measure) operator once you have your word vector created and your label defined.

SilkeKoch

Thank you very much for your help. I tried both. If there is any tutorial, book, text etc to understand the "values" of the weights better, let me know
Many thanks!