Reproduce “Hypergraph-Enhanced Evidence Selection for Long-Context Multimodal Document QA” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Wenxin Zhao, Jiachuan Wang, Xinyi Zhu, Jiake Ge, Haoyang Li, Yongqi Zhang, Shuangyin Li, Lei Chen
HyperDoc selects related evidence groups from long multimodal documents rather than retrieving isolated pages. It links visual anchors with captions and related text, then chooses groups according to query scope and input cost. On MMLongBench-Doc with Qwen3-VL 8B, it reports 53.2% generalized ACC versus 32.8% for MDocAgent; removing contextual hyperedges lowers ACC to 43.0%, but does not isolate grouping alone.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.