Predicting the Popularity of Trending Arabic News on Twitter
Nourh Abdulaziz Rsheed, Muhammad Badruddin Khan · 2014
Twitter is a popular form of social media that helps to spread breaking news. Mining textual documents and the number of posts ("tweets") can assist in predicting the popularity of news articles based on the content of news articles. The purpose of this study is to build a model that can predict the popularity of news articles on Twitter by classifying their features, providing comparisons using three algorithms for data mining: decision tree, NB, and rule based W-JRIP. Four approaches were used to compare the extracted features: light stemming, N-grams and light stemming, N-grams, and bag of words. The results of the experiment on the external features of articles indicate that the decision tree performed better (93.30%) than the rule-based and NB algorithms. The experiment on the internal features yielded no significant predictors of the popularity of news articles on Twitter. The results of the experiments on internal and external features of articles indicate that the decision tree algorithm with light stemming approaches performs best, achieving 85.60%. This study expands the limited literature on Arabic text classification and provides a predictive model that may help news organizations improve their online contents.