AI RESEARCH
BRIDGE: Multimodal-to-Text Retrieval via Reinforcement-Learned Query Alignment
arXiv CS.CV
•
ArXi:2604.07201v1 Announce Type: cross Multimodal retrieval systems struggle to resolve image-text queries against text-only corpora: the best vision-language encoder achieves only 27.6 nDCG on MM-BRIGHT, underperforming strong text-only retrievers. We argue the bottleneck is not the retriever but the query -- raw multimodal queries entangle visual descriptions, conversational noise, and retrieval intent in ways that systematically degrade embedding similarity. We present \textbf{BRIDGE}, a two-component system that resolves this mismatch without multimodal encoders.