AmbigChat: Interactive Hierarchical Clarification for Ambiguous Open-Domain Question Answering
summary
AmbigChat constructs an automated disambiguation tree by combining an ambiguity taxonomy from 5,000 queries with web search and LLM grounding, extracting 10 facet types and hierarchically generating facet values (each facet limited to ~2–5 options). The interface surfaces interactive clarification widgets, supports drag-and-drop context chips, and adds new branches on demand. A design process with six professionals precedes a user study with 16 participants that reports faster, easier answer finding and structured exploration of the answer space.
AI-written summary · about
chapters (6)
- 0:00Introduction
- 0:24Ambiguity in questions
- 3:53Disambiguation tree
- 5:38User interaction
- 8:51Exploration examples
- 9:58Conclusion and Q&A
related talks
15mRAG Beyond Chatbots
29mThe GraphRAG System That Asks Back: Intelligent Probing with Contextual Answers
13mInteractive Learning and User‑Centered AI: A Retrospective
10mFrom Philosophy to Interfaces: an Explanatory Method and a Tool Inspired by Achinstein's Theory ...
16mUser Interaction Models for Disambiguation in Programming by Example
22mAll Work and No Play? Conversations with a Question-and-Answer Chatbot in the Wild
20mConvey: Exploring the Use of a Context View for Chatbots
09mEnhance Healthcare Chatbot Transparency