A Dataset for Authorship Analysis of Short Modern Arabic Text

Abstract

The collection has 71391 Arabic tweets written by 44 Arab author from 13 Arab Countries plus tweets of Pope Francis in MSA. Tweepy API was used for tweet scraping. Tweet topics: Politics, Journalism, Religion, Arabic Literature. The tweets were preprocessed: Retweets and replies were removed, any tweet with the author’s name was removed, emojis, hyperlinks, and English hashtags were filtered out, numbers were normalized and English and other non-Arabic characters were filtered out. Then stop words were removed, tokens lemmatized and normalized. The two files have the training and testing data used in our project.