Evaluating the Effectiveness of Visualization Design for Twitter Conversations on Academic Topics

Abstract

Our research focuses on evaluating the efficiency of the combination of the computed scores and visual representations of features. W conducted a two-phased user study. The first phase were designed to determine the important text features. In the first phase, participants were asked to rate a set of Twitter data and prominent features. After the key feature set was determined, we started to evaluate the two sources of results to be visualized: 1) informativeness and relatedness scores and 2) visual features. We built four versions of visualizations to represent the full projection of the two sources. In the second phase of the user study, 48 participants were recruited to do a between-subject user study for evaluating the four versions of visualizations using the same set of information seeking tasks. The study shows that the visualization combining scores and features perform the best in efficiency for browsing tasks.