Comparative Analysis of Statistical Classifiers for Predicting News Popularity on Social Web

摘要

In recent years, owing mainly to the ubiquity of the Internet, social media platforms are more popular than ever before. Facebook boasts of over 2 billion registered accounts and the number of individual users is said to be more than the population of most countries. It is clear that social media has our attention, and media houses are no strangers to this fact. A huge amount of time and resource is put into the research and development of strategies that will help news flash become more popular. One of the major driving factors of news popularity is the sentiment and emotion behind the news. Human emotions are the driving force of any microblog on social media today and in our research, we attempt to study some of these fields that affect the mood of people. These features include specific properties about the news, such as the sentiment and the topic of the news itself. They further include factors unrelated to the news articles that may affect the news reading behavior of readers, like the day of week or time of the day. Our research provides an approach to design a predictive model for the popularity of a news article on a particular social media platform, based on the input features.