Asking the Missing Piece: Context-Driven Clarification for Ambiguous VQA
Dec 1, 2025·
,·
1 min read
Zhen Cao
Bingbing Wen
Lucy Lu Wang
Abstract
We study context-driven clarification strategies for ambiguous visual question answering, enabling models to ask targeted follow-up queries when the available context is insufficient to answer reliably.
Type
Publication
NeurIPS 2025 Workshop on Foundations of Reasoning in Language Models
We explore how VQA systems can ask targeted clarification questions when initial context is ambiguous, improving reliability and interpretability in multimodal reasoning.
