Identifying the drivers of negative news with sentiment, entity and regression analysis

Abstract

Modern-day news agencies cater for a wide range of negative news, since multiple studies show general people are more attracted towards negative news. Once a highly negative incident is reported by a local news agency, it is often propagated by many other foreign news agencies at a global scale characterizing the news as breaking news. This propagation of negative news generates significant impacts on groups (who conducted the event), location (where the event was conducted), societies (that was impacted by the news) along with other factors. This research critically analyzes the impacts of negative news or breaking news with the help Artificial Intelligence (AI) based techniques like sentiment analysis, entity detection and automated regression analysis. The methodology described within this paper was implemented with a unique algorithm that allowed identification of all related factors or topics that drive negative perceptions towards global news. The solution was hosted in cloud environment from 2nd June 2021 till 1st September 2021. It automatically captured and analyzed 22,425 global news from 2397 different news sources of 192 countries. During this time, 34,975 entities are automatically categorized into 13 different entity groups. The classification accuracy of the entity detection was found to be 0.992, 0.995 and 0.994 in terms of precision, recall and F1-score. Moreover, the accuracies of logistic regression and linear regression were found to be 0.895 in AUC and 0.255 in MAPE on an average. Finally, the presented solution was successfully deployed in a wide range of environments including smartphones, tablets, and desktops.