An Interactive Visual System for Data Analytics of Social Media

With the development of the Internet, Big Data analysis of social media has become a hot research topic in recent years. However, a challenging problem in social media analysis is how to intelligently detect event information from massive media data and how to help users quickly understand event content. Therefore, we propose a method to extract important time periods of events from media data to analyze the media event information, and develop an interactive visual system based on the "5W" principle. The system provides an interactive analysis platform for exploring media Big Data (e.g., Twitter data). The system uses a dynamic topic model to extract topics, a Naive Bayesian classifier to distinguish emotions, and explores events based on the evolution of time, topics, and emotions. In terms of visualization, the system allows users to add some annotations based on segmentation of complex information and highlight important events by adding different story cards to the timeline so that users can get a quick overview of events. Users can also interactively modify the story cards during the exploration process. Finally, several case studies show that our system can effectively reduce the time required to understand social media data and allow users to quickly explore the full picture of an event through an interactive approach.