ViLBias：检测和推理解多模态内容中的偏见
ViLBias: Detecting and Reasoning about Bias in Multimodal Content
AI总结：
ViLBias通过多模态基准检测新闻中的偏见，利用LLM和VLMs提升推理准确率和忠实度，参数高效方法能恢复大部分微调性能。
摘要：
检测多模态新闻中的偏见需要能够对文本-图像对进行推理的模型，而不仅仅是对文本进行分类。为此，我们提出了ViLBias，一种类似于VQA的基准和框架，用于检测和推理解多模态新闻中的偏见。该数据集包含来自不同来源的40,945个文本-图像对，每个都带有偏见标签和简要的解释，使用两阶段的LLM作为标注器流程，结合层次多数投票和人机交互验证。我们评估了小型语言模型（SLMs）、大型语言模型（LLMs）和视觉-语言模型（VLMs）在封闭式分类和开放式推理（oVQA）上的表现，并比较了参数高效微调策略。结果表明，结合图像和文本可以提高检测准确率3-5%，并且LLM/VLMs比SLMs更能捕捉到细微的框架和文本-图像不一致之处。参数高效方法（LoRA/QLoRA/Adapters）可以恢复97-99%的全微调性能，仅需<5%的可训练参数。对于oVQA，推理准确率在52-79%之间，忠实度在68-89%之间，均通过指令微调得到提升；封闭式准确率与推理能力（r=0.91）呈强相关。ViLBias提供了一个可扩展的基准和强大的基线，用于多模态偏见检测和解释质量。
英文摘要：
Detecting bias in multimodal news requires models that reason over text--image pairs, not just classify text. In response, we present ViLBias, a VQA-style benchmark and framework for detecting and reasoning about bias in multimodal news. The dataset comprises 40,945 text--image pairs from diverse outlets, each annotated with a bias label and concise rationale using a two-stage LLM-as-annotator pipeline with hierarchical majority voting and human-in-the-loop validation. We evaluate Small Language Models (SLMs), Large Language Models (LLMs), and Vision--Language Models (VLMs) across closed-ended classification and open-ended reasoning (oVQA), and compare parameter-efficient tuning strategies. Results show that incorporating images alongside text improves detection accuracy by 3--5\%, and that LLMs/VLMs better capture subtle framing and text--image inconsistencies than SLMs. Parameter-efficient methods (LoRA/QLoRA/Adapters) recover 97--99\% of full fine-tuning performance with $<5\%$ trainable parameters. For oVQA, reasoning accuracy spans 52--79\% and faithfulness 68--89\%, both improved by instruction tuning; closed accuracy correlates strongly with reasoning ($r = 0.91$). ViLBias offers a scalable benchmark and strong baselines for multimodal bias detection and rationale quality.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
