Visualizing Large Volumes of Feedback for Designers

Abstract

Reviewing feedback is an important part of the creative process, allowing designers to determine the effectiveness of their work and make critical decisions. This practice becomes challenging when feedback is collected in high volumes. Without a way to organize, visualize, and explore data, such feedback is unusable. This research presents the design for an interface which automatically categorizes large volumes of feedback through machine learning, generates an interactive visualization which progressively discloses data to users, and provides tools for navigating and manipulating feedback, allowing designers to uncover and record insights. Eight professional designers were interviewed to determine the functionality of this interface and later took part in user tests to judge its effectiveness. User tests confirmed that the automatic categorization, interactive visualizations, navigation features, and data manipulation tools were largely helpful for designers. Results also revealed areas for further development, some of which include empowering users to train the machine learning, evolving the visualizations to show how data overlaps and changes over time, and increasing transparency so that users know what information was organized by a human versus machine learning.