Retrieval enhancement generation method and system based on hybrid retrieval and self-adaptive sorting
By employing a hybrid retrieval and adaptive ranking approach, the problem of limited retrieval methods and rigid ranking strategies in existing RAG systems is solved, achieving high relevance and accuracy of generated content, and making it suitable for various application scenarios such as intelligent customer service and knowledge Q&A.
CN121636677APending Publication Date: 2026-03-10INSPUR ZHUOSHU BIG DATA IND DEV CO LTD
0 Cites 8 Cited by
Patent Information
- Application Number
- CN202511728668.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-24
- Publication Date
- 2026-03-10
AI Technical Summary
Technical Problem
The existing RAG system suffers from a single retrieval method and a rigid sorting strategy, resulting in factual errors, information lag, and mismatches with generation requirements in the generated content.
Method used
A hybrid retrieval mechanism is adopted to integrate dense and sparse retrieval, and an adaptive sorting module is designed to dynamically adjust the sorting weights, thereby improving the accuracy of candidate document screening and sorting.
Benefits of technology
It significantly improves the relevance, accuracy, and contextual consistency of generated content, reduces illusions and missing information, and is suitable for a variety of application scenarios.
✦ Generated by Eureka AI based on patent content.
Abstract
The invention discloses a retrieval enhancement generation method and system based on hybrid retrieval and adaptive sorting, and relates to the technical field of artificial intelligence and natural language processing. Comprising the following steps: 1, analyzing a query language and providing multi-path retrieval: receiving a natural language query input by a user, and performing semantic analysis and structured processing; starting dense vector retrieval and sparse semantic retrieval in parallel, and respectively obtaining candidate document sets from the knowledge base; 2, candidate mixed result fusion is carried out, wherein duplicate removal and preliminary fusion are carried out on candidate documents obtained through dense retrieval and sparse retrieval, and a unified candidate document pool is formed; all the candidate documents are evaluated according to the query semantic matching degree, document authority and quality, context coherence and generation task type factors, sorting weights are dynamically generated based on all the factors, and a candidate document pool is resorted; and constructing a structured context prompt prompt, inputting the structured context prompt prompt into a pre-trained large model, and generating final response content.
Need to check novelty before this filing date? Find Prior Art