Trustworthy AI

ArgVLLMs: Structuring Multimodal Reasoning with Quantitative Bipolar Argumentation for Faithful and Contestable Explainability