Familiarity Vs Trust: A Comparative Study of Domain Scientists' Trust in Visual Analytics and Conventional Analysis...

Area of Science:

• Computer Science
• Biology
• Human-Computer Interaction

Background:

• Interactive visualization and automated analysis drive data-driven discovery.
• Mixed-initiative systems enable human-machine collaboration in analysis.
• Trust in machine-generated outputs is a key research question for domain experts.

Purpose of the Study:

• To evaluate trust in visual analytics systems among biologists.
• To investigate the impact of familiarity and task complexity on trust.
• To propose a trust-augmented design for visual analytics.

Main Methods:

• Designed and evaluated a mixed-initiative visual analytics system for biologists.
• Conducted a controlled user study with 34 biologists.
• Compared trust levels between conventional and visual analytic mediums.

Main Results:

• Biologists reported comparable trust in unfamiliar visual analytics as in conventional methods.
• Visual analytics systems inspired greater trust for complex sense-making tasks.
• Familiarity did not significantly decrease trust in visual analytics.

Conclusions:

• Visual analytics can be a trustworthy analysis medium for biologists.
• Trust in visual analytics is influenced by task complexity.
• Future research should focus on enhancing the trustworthiness of visual analytic systems.