Reproduce “From Facts to Foils: Designing and Evaluating Counterfactual Explanations for Smart Environments” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Anna Trapp, Mersedeh Sadeghi, Andreas Vogelsang

This research defines and implements counterfactual explanations for rule-based smart environments, offering a framework to generate actionable "what-if" scenarios. A user study demonstrates that preferences for causal versus counterfactual explanations are context-dependent, influenced by factors like desired actionability and time pressure.

Key insights

  • Counterfactual explanations effectively provide actionable advice, allowing users to understand how to achieve desired system outcomes, thereby addressing a gap left by purely causal explanations.
  • The utility and user preference for explanations are highly context-dependent, varying with linguistic clarity, content usefulness, user goals, and situational factors like time pressure or the desire to change an outcome.
  • A trade-off exists between the linguistic simplicity of causal explanations and the actionable value embedded in counterfactuals, suggesting the need for context-adaptive XAI systems.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.