印度政府监管文档RAG中解析、分块与嵌入的交互研究
原标题：Parser, Chunking, and Embedding Interactions in Retrieval-Augmented Generation over Indian Government Regulatory Documents
AI 摘要
该研究对印度政府监管文档上的RAG流水线进行了受控因子实验，联合评估解析器、分块策略和稠密嵌入模型，并与BM25稀疏基线对比。结果显示：没有单一检索器家族在所有文档上占优；解析器与分块器选择存在显著交互，最佳组合无法从单个组件排名预测；MPNet-base嵌入模型表现持续较差，在表格类问题上失败严重；语料证据保留接近饱和，检索差异主要来自排序质量而非信息损失。作者发布了评估工具、语料清单和800题基准。
正文节选
Parser, Chunking, and Embedding Interactions in Retrieval-Augmented Generation over Indian Government Regulatory Documents Abstract Retrieval-augmented generation (RAG) pipelines are typically assembled from independently-chosen components – a document parser, a chunking strategy, and an embedding model – yet these choices are rarely evaluated jointly, and evaluations that do combine them are usually run on a single document or corpus. We present a controlled factorial study of parsers chunking