User Interaction Dataset for JAIR 2024 paper "Opening the Analogical Portal to Explainability: Can Analogies Help Laypeople in AI-assisted Decision Making?"

This repo contains all code and data for the paper "Opening the Analogical Portal to Explainability: Can Analogies Help Laypeople in AI-assisted Decision Making?" We collected user decision making with AI assistants in a skin cancer detection scenario. Specifically, we provide concept-level explanations generated with post-hoc XAI methods and analogy-based explanations to elucidate the importance of concepts. User interfaces built upon flask is also provided.

本代码仓库包含论文《打开可解释性的类比之门：类比能否助力非专业人士开展AI辅助决策？》（Opening the Analogical Portal to Explainability: Can Analogies Help Laypeople in AI-assisted Decision Making?）的全部代码与数据集。我们在皮肤癌检测场景中收集了用户借助AI助手开展决策的相关数据。具体而言，本仓库提供了基于事后可解释AI（post-hoc XAI）方法生成的概念级解释，以及基于类比的解释，用以阐明各类概念的重要性。同时还提供了基于Flask框架构建的用户界面。