Comparison Conundrum and the Chamber of Visualizations: An Exploration of How Language Influences Visual Design

Area of Science:

• Human-Computer Interaction
• Data Visualization
• Natural Language Processing

Background:

• Supporting natural language-based visual comparison is challenging due to complex language.
• Understanding user reasoning in natural language comparisons is crucial for interface design.

Purpose of the Study:

• To explore the design space of utterances for comparing data entities.
• To investigate how users design visualizations for comparisons based on natural language.
• To identify guidelines for natural language interfaces supporting visual comparisons.

Main Methods:

• Conducted a user study with visualization experts and non-experts to explore comparison utterance design space.
• Extracted key design features from user-created visualizations.
• Synthesized features into sixteen representative visualization designs.
• Validated user preferences for these designs in a follow-up study.

Main Results:

• Identified parameters for comparison utterances (data variables, attributes) and their specification (explicit/implicit).
• Developed sixteen representative visualization designs based on user studies.
• Validated user preferences for specific visualization designs corresponding to natural language utterances.

Conclusions:

• Findings provide insights into user reasoning for visual comparisons.
• Suggests guidelines for designing natural language interfaces and recommendation tools.
• Highlights future directions for supporting natural language comparisons in visual analytics.