作者位次 #3/6SIGIR 2026Oral· Information Retrieval (Full)
RegionSLM: Region-aware Question Answering on Document Screenshots
Chao Wang · Hehe Fan · Huichen Yang · Sarvnaz Karimi · Lina Yao · Yi Yang
摘要
Real-world document question-answering that relies on screenshots, such as bills and forms, requires evidence that is often spatially localised and visually cluttered. However, most Screenshot Language Models (SLMs) encode the entire page holistically and rely on implicit attention to ''find'' re...