MultiPress: A Multi-Agent Framework for Interpretable Multimodal News Classification
This work addresses the challenge of effectively classifying multimodal news content for applications in media and information processing, representing a novel method for a known bottleneck.
The paper tackles the problem of multimodal news classification by proposing MultiPress, a three-stage multi-agent framework that integrates specialized agents for perception, reasoning, and fusion, achieving significant improvements in classification accuracy and interpretability on a new large-scale dataset.
With the growing prevalence of multimodal news content, effective news topic classification demands models capable of jointly understanding and reasoning over heterogeneous data such as text and images. Existing methods often process modalities independently or employ simplistic fusion strategies, limiting their ability to capture complex cross-modal interactions and leverage external knowledge. To overcome these limitations, we propose MultiPress, a novel three-stage multi-agent framework for multimodal news classification. MultiPress integrates specialized agents for multimodal perception, retrieval-augmented reasoning, and gated fusion scoring, followed by a reward-driven iterative optimization mechanism. We validate MultiPress on a newly constructed large-scale multimodal news dataset, demonstrating significant improvements over strong baselines and highlighting the effectiveness of modular multi-agent collaboration and retrieval-augmented reasoning in enhancing classification accuracy and interpretability.