Big Social Data Analytics for Public Health: Predicting Facebook Post Performance Using Artificial Neural Networks and Deep Learning
Nadiya Straton, Raghava Rao Mukkamala, Ravi Vatrapu · 2017
Facebook "post popularity" analysis is fundamental for differentiating between relevant posts and posts with low user engagement and consequently their characteristics. This research study aims at health and care organizations to improve information dissemination on social media platforms by reducing clutter and noise. At the same time, it will help users navigate through vast amount of information in direction of the relevant health and care content. Furthermore, study explores prediction of popularity of healthcare posts on the largest social media platform Facebook. Methodology is presented in this paper to predict user engagement based on eleven characteristics of the post: Post Type, Hour Span, Facebook Wall Category, Level, Country, isHoliday, Season, Created Year, Month, Day of the Week, Time of the Day. Finally, post performance prediction is conducted using Artificial Neural Networks (ANN) and Deep Neural Networks (DNN). Different network topology measures are used to achieve best accuracy prediction followed by examples and discussion on why DNN might not be optimal technique for the given data set.