A Conversational Assistant for Democratization of Data Visualization: A Comparative Study of Two Approaches of Interaction

ABSTRACT

Data visualization is an important aspect of exploratory data analysis which requires expertise, posing a challenge to many users. In this work, we present a conversational system designed to make data visualization accessible, where users can generate visualizations of data by natural language conversational interaction with the system. We conducted a between‐subject study to compare two versions of our system, where users were asked to perform data exploration tasks. We did not find significant differences as concerns user assessments of their interactions with the system; however, we uncovered that users explore different properties of the data and reach different conclusions when using the two systems. We will discuss the general implications of our findings as regards the usage of conversational interfaces for visualization.