用图结构模拟人类思考，让AI回答更符合人类偏好。
A Comprehensive Graph Framework for Question Answering with Mode-Seeking Preference Alignment
- 构建分层文档图，模仿人脑理解信息的方式。
- 通过概率匹配优化，使答案更贴近人类偏好。
- 在6个数据集上验证，显著提升回答质量。
近期检索增强生成（RAG）的发展提升了大模型在问答任务中的表现，但全局理解与人类伦理及质量偏好对齐仍存挑战。为此，我们提出GraphMPA——一种基于图结构的综合框架，具备模式寻找偏好对齐能力。该方法利用通用相似性度量构建分层文档图，模拟人类认知过程以实现信息理解与整合；同时引入模式寻找偏好优化，通过概率匹配约束使模型输出更符合人类偏好。在六个数据集上的大量实验验证了GraphMPA的有效性。
原文摘要 · Abstract (English)
Recent advancements in retrieval-augmented generation (RAG) have enhanced large language models in question answering by integrating external knowledge. However, challenges persist in achieving global understanding and aligning responses with human ethical and quality preferences. To address these issues, we propose GraphMPA, a comprehensive graph-based framework with mode-seeking preference alignment. Our approach constructs a hierarchical document graph using a general similarity measurement, mimicking human cognitive processes for information understanding and synthesis. Additionally, we introduce mode-seeking preference optimization to better align model outputs with human preferences through probability-matching constraints. Extensive experiments on six datasets demonstrate the effectiveness of our \href{https://github.com/tangquanwei/GraphMPA}{GraphMPA}.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。