MARQUIS: A Three-Stage Pipeline for Video Retrieval-Augmented Generation
This work addresses the dual challenge of complex video retrieval and coherent multi-video synthesis for automated article generation, achieving large gains on a shared task benchmark.
MARQUIS introduces a three-stage pipeline for video retrieval-augmented generation that improves retrieval nDCG@10 from 0.195 to 0.759 and boosts human evaluation scores for article generation from 3.09 to 3.83 over the CAG baseline.
Retrieval-augmented generation from videos requires systems to retrieve relevant audiovisual evidence from large corpora and synthesize it into coherent, attributed text. Current approaches struggle at both ends: retrieval methods fail on complex, multi-faceted queries that cannot be captured by a single embedding, while generation methods lack the high-level reasoning needed to synthesize across multiple videos and face memory constraints over long, multi-video contexts. We present MARQUIS: a three-stage pipeline that addresses these limitations through (1) query expansion, fusion, and reranking, (2) calibrated structured evidence extraction, and (3) article generation from extracted evidence, optionally controlled by an RLM. On the MAGMaR2026 shared task, we improve retrieval performance from 0.195 to 0.759 (nDCG@10). For article generation, ITER-QA-BASE improves average human score from 3.09 to 3.83 over the CAG baseline, while MARQUIS-RLM achieves a human score of 3.30 and the strongest citation recall among non-QA systems.