TrustME: A Context-Aware Explainability Model to Promote User Trust in Guidance

Area of Science:

• Human-Computer Interaction
• Information Visualization
• Explainable AI

Background:

• Guidance-enhanced visual analytics (VA) aids users in data interpretation.
• Lack of clear explanations for system interventions can reduce guidance effectiveness.
• Effective guidance requires context-specific explanations to build user trust.

Purpose of the Study:

• To introduce the concept of explainable guidance in VA.
• To propose a model for designing explainability strategies for VA guidance.
• To investigate the impact of explainable guidance on user trust and system adoption.

Main Methods:

• Literature review on explainable AI, VA guidance, and user-system trust.
• Development of a model for designing explainable guidance strategies.
• Application of the model to two VA scenarios and a design walk-through with an expert.

Main Results:

• The proposed model facilitates the design of context-aware explanations for VA guidance.
• Explainable guidance positively influences user trust in VA systems.
• The model aids designers in articulating the rationale behind system interventions.

Conclusions:

• Explainable guidance is essential for effective VA systems.
• The proposed model provides a framework for designing trustworthy and context-effective guidance.
• Future research should explore the empirical validation of explainable guidance strategies in diverse VA contexts.