用混合检索与验证机制，让大模型回答企业文档问题时更准、可追溯、安全可靠。
eSapiens's DEREK Module: Deep Extraction & Reasoning Engine for Knowledge with LLMs
- 分块+混合向量+关键词检索，提升文档问答召回率和精度
- 验证器确保每条答案有引用支撑，错误陈述少于3%
- 适合法律金融等高要求场景，支持端到端加密与容器部署
我们提出 DEREK（Deep Extraction & Reasoning Engine for Knowledge）模块，一种专为企事业单位文档问答设计的安全可扩展检索增强生成系统。该系统由 eSapiens 设计实现，能处理 PDF、Office、网页等多种格式内容，将其分割为 1000 字符重叠块，并存入混合的 HNSW+BM25 索引中。用户查询经 GPT-4o 优化后，通过向量与 BM25 联合搜索获取结果，再由 Cohere 重排序，最终由大模型基于 CO-STAR 提示工程生成答案。一个 LangGraph 验证器强制要求每个主张都有引用重叠，不满足则重新生成，直至符合要求。在 LegalBench 的四个子集上，使用 1000 字符块使 Recall@50 提升约 1 个百分点，混合检索+重排序使 Precision@10 提升约 7 个百分点；验证器使 TRACe Utilization 超过 0.50，且未支持陈述占比低于 3%。所有组件均运行在容器中，支持端到端 TLS 1.3 和 AES-256 加密。实验表明，DEREK 模块可在极低运维成本下提供准确、可追溯、生产可用的文档问答能力，满足法律与金融等高风险领域的安全审计需求。
原文摘要 · Abstract (English)
We present the DEREK (Deep Extraction & Reasoning Engine for Knowledge) Module, a secure and scalable Retrieval-Augmented Generation pipeline designed specifically for enterprise document question answering. Designed and implemented by eSapiens, the system ingests heterogeneous content (PDF, Office, web), splits it into 1,000-token overlapping chunks, and indexes them in a hybrid HNSW+BM25 store. User queries are refined by GPT-4o, retrieved via combined vector+BM25 search, reranked with Cohere, and answered by an LLM using CO-STAR prompt engineering. A LangGraph verifier enforces citation overlap, regenerating answers until every claim is grounded. On four LegalBench subsets, 1000-token chunks improve Recall@50 by approximately 1 pp and hybrid+rerank boosts Precision@10 by approximately 7 pp; the verifier raises TRACe Utilization above 0.50 and limits unsupported statements to less than 3%. All components run in containers, enforce end-to-end TLS 1.3 and AES-256. These results demonstrate that the DEREK module delivers accurate, traceable, and production-ready document QA with minimal operational overhead. The module is designed to meet enterprise demands for secure, auditable, and context-faithful retrieval, providing a reliable baseline for high-stakes domains such as legal and finance.
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。