VARTTA: A Visual Analytics System for Making Sense of Real-Time Twitter Data

Amir Haghighati,Kamran Sedig

doi:10.3390/data5010020

Abstract

Through social media platforms, massive amounts of data are being produced. As a microblogging social media platform, Twitter enables its users to post short updates as “tweets” on an unprecedented scale. Once analyzed using machine learning (ML) techniques and in aggregate, Twitter data can be an invaluable resource for gaining insight into different domains of discussion and public opinion. However, when applied to real-time data streams, due to covariate shifts in the data (i.e., changes in the distributions of the inputs of ML algorithms), existing ML approaches result in different types of biases and provide uncertain outputs. In this paper, we describe VARTTA (Visual Analytics for Real-Time Twitter datA), a visual analytics system that combines data visualizations, human-data interaction, and ML algorithms to help users monitor, analyze, and make sense of the streams of tweets in a real-time manner. As a case study, we demonstrate the use of VARTTA in political discussions. VARTTA not only provides users with powerful analytical tools, but also enables them to diagnose and to heuristically suggest fixes for the errors in the outcome, resulting in a more detailed understanding of the tweets. Finally, we outline several issues to be considered while designing other similar visual analytics systems.

Highlights

IntroductionSocial media platforms allow their users to simultaneously be active consumers (readers) as well as producers (authors or editors) of data
Social media platforms allow their users to simultaneously be active consumers as well as producers of data
We demonstrate the use of VARTTA in political discussions

Summary

Introduction

Social media platforms allow their users to simultaneously be active consumers (readers) as well as producers (authors or editors) of data. Pendry et al [4] argue that social platforms have benefits at both individual and societal levels, and that due to the richness of their content, they are of greater applied importance than has been realized Depending on their features, user-produced content on microblogging platforms may include various combinations of data types (e.g., text, videos, images, and hyperlinks). Assessing validity of arguments, discerning most and least significant contributors, understanding sequence and network of communications, identifying themes and intentions of tweets, and reviewing identities and backgrounds of contributors are examples where one or both of these approaches can be used to enhance the outcome. The role of tweeters in any given group can further expose hidden aspects of expressions made by them which directly relates to “why”

Objectives

Methods

Conclusion