VISTX: Experimenting with Interactive Stance and Sentiment Visualisation
Katherine Grayson, Jon Chamberlain · 2019
User-generated content online continues to grow rapidly, and with this it becomes more challenging to evaluate the general consensus on a trending topic. Even greater challenges are faced if one wishes to compare opinions across multiple topics, as it can be difficult to make comparisons and reach conclusions with large amounts of data in an intuitive manner. VISTX explores innovative methods of visualising opinion in terms of stance and sentiment using Twitter data from the SemEval-2016 Stance Dataset and presents three types of interactive visualisations. We discuss the method of testing the visualisations for effectiveness with a small group of users, where time taken and the number of incorrect interpretations were measured for each user. We discuss the results, which indicate that the `circle-packing' visualisation type may lead to quicker interpretations for five or more topics. Finally, we discuss the work needed to further solidify these initial findings.