视觉语言模型在UML图解释中的先验偏差
Prior Bias in Vision Language Models on UML Diagram Interpretation
AI总结：
研究视觉语言模型在UML图解释时是依据先验知识还是真正理解图表。通过引入对比基准测试，评估多个模型，发现反转关系箭头会降低开源模型关系方向准确率，不同模型表现有差异，揭示先验驱动的理解故障。
摘要：
视觉语言模型（VLMs）越来越多地应用于软件工程工件，特别是其含义取决于视觉符号的UML类图。然而，尚不清楚VLMs是真正读取此类图，还是从关于类通常如何关联的预训练先验中回答。我们引入了一个受控的UML基准，其中每个符合先验的图都与其先验冲突的对应图配对，该对应图（1）保留相同的类名和布局，同时（2）仅反转关系箭头。我们评估了来自两个模型家族InternVL3.5和Qwen3的八个开源VLMs，以及两个闭源前沿模型GPT-5.4和GPT-5.4 Mini。在八个开源模型中，反转箭头平均会使关系方向准确率降低33.48%，而GPT-5.4 Mini保持10%的差距。在更难的三类条件下，开源模型的准确率平均大幅下降45.28%，对于GPT-5.4家族甚至下降18.62%。扩展仅提供有限的改进，并且依赖于家族。我们的基准展示了基于图的软件理解中由先验驱动的诊断故障。我们的工件可在此https URL获得。
英文摘要：
Vision Language Models (VLMs) are increasingly applied to software engineering artifacts, especially UML class diagrams whose meaning depends on visual notation. Yet, it is unclear whether VLMs actually read such diagrams or instead answer from pretrained priors about how classes typically relate. We introduce a controlled UML benchmark in which each prior-conforming diagram is paired with its prior-conflicting counterpart that (1) preserves the same class names and layout while (2) reverses only the relation arrow. We evaluate eight open-source VLMs from two model families, InternVL3.5 and Qwen3, alongside two closed-source frontier models GPT-5.4 and GPT-5.4 Mini. Across the eight open-source models, reversing the arrow reduces relation-direction accuracy by 33.48% on average, while GPT-5.4 Mini retains a 10% gap. In the harder three-class condition, accuracy drops sharply by 45.28% for open-source models, and even 18.62% for the GPT-5.4 family on average. Scaling provides only limited improvements and is family-dependent. Our benchmark presents a diagnostic prior-driven failure in diagram-grounded software understanding. Our artifact is available at https://anonymous.4open.science/r/UMLKnowledgeConflict-8461.
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。
