2
$\begingroup$

I got a dataset made out of tweets and I need to classify them into topics. For topic modelling with LDA I have cleaned out the dataset (removing stopwords, mentions, symbols, etc). Do I need to remove also retweets in order not to influence the LDA? If so, how should I finally classify all tweets since I could not use per-document-per-topic probability?

$\endgroup$
0