MAGMaR 2026 is the second workshop focused on multimodal augmented generation via multimodal retrieval, building on the inaugural ACL 2025 workshop. It aims to advance research in multimodal content retrieval (text, image, audio, video) and retrieval-augmented generation, with a shared task on generating grounded articles from multiple videos to combat misinformation and support journalistic synthesis.
A strong submission should clearly identify its contribution and evaluate it appropriately.
Research Papers
Novel research papers on topics including retrieval-augmented generation, multimodal understanding, multimodal retrieval, document retrieval, multimodal question answering, and video/image/audio understanding. Both archival and non-archival submissions are permitted.
Shared Task System Papers
System description papers from teams participating in the retrieval or generation tracks of the shared task, detailing their approaches.