Counterfactual Explanations for Natural Language Interfaces
George Tolkachev University of Pennsylvania \AndStephen Mell University of Pennsylvania \ANDSteve Zdancewic University of Pennsylvania \AndOsbert Bastani University of Pennsylvania
Abstract
A key challenge facing natural language interfaces is enabling users to understand the capabilities of the underlying system. We propose a novel approach for generating explanations of a natural language interface based on semantic parsing. We focus on counterfactual explanations, which are post-hoc explanations that describe to the user how they could have minimally modified their utterance to achieve their desired goal. In particular, the user provides an utterance along with a demonstration of their desired goal; then, our algorithm synthesizes a paraphrase of their utterance that is guaranteed to achieve their goal. In two user studies, we demonstrate that our approach substantially improves user performance, and that it generates explanations that more closely match the user’s intent compared to two ablations.111Code available at: https://github.com/georgeto20/counterfactual_explanations.
中文速览
自然语言接口（natural language interface）常常听不懂用户的指令，因为用户的措辞可能偏离系统能理解的表达范围，而用户自己却不清楚症结在哪。研究者提出了一种基于语义解析（semantic parsing）的反事实解释（counterfactual explanation）算法：当系统无法正确执行用户指令时，用户只需额外提供一段演示来说明自己的真实意图，算法便会在系统可识别的语法范围内，自动找到一条与原始指令语义最接近、同时又保证能被正确执行的改写表达。研究者在 BabyAI 虚拟导航环境中开展了两项用户实验，结果显示该方法生成的解释比两种对照基线更忠实于用户原意，并且显著提升了用户后续提供有效指令的成功率。这项工作为让普通用户"摸清"AI 系统的能力边界提供了一条实用路径，对提升人机交互的透明度和可用性具有重要意义。
原文 arXiv:2204.13192;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2204.13192v1