Enhanced search generation method and apparatus, electronic device, and storage medium

By representing financial events in a hypergraph structure and using hyperedges and subgraph matching templates, the problem that traditional graph structures cannot fully express complex events involving multiple entities is solved, thus enabling more efficient financial information retrieval.

CN121501854BActive Publication Date: 2026-07-10IFLYTEK CO LTD
View PDF 3 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
IFLYTEK CO LTD
Filing Date
2026-01-13
Publication Date
2026-07-10

AI Technical Summary

Technical Problem

Traditional graph structures cannot fully represent complex events involving multiple entities when representing financial events, resulting in low retrieval efficiency and incomplete recall, failing to meet the requirements of accuracy, traceability, and auditability in financial scenarios.

Method used

A hypergraph structure is used to represent composite events in the form of hyperedges. Isomorphic matching is performed in the hypergraph structure through subgraph matching templates. The response text is generated by combining a natural language model, while preserving paragraph-level semantic context.

Benefits of technology

It achieves accurate expression of complex financial semantics, improves the accuracy and reliability of queries, and meets the requirements of credibility and auditability in financial scenarios.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121501854B_ABST
    Figure CN121501854B_ABST
Patent Text Reader

Abstract

The application discloses an enhanced retrieval generation method and device, electronic equipment and storage medium, and belongs to the technical field of artificial intelligence. The method comprises the following steps: receiving a query sentence input by a user, and parsing the query sentence to obtain a subgraph matching template; performing subgraph isomorphism matching on a hypergraph structure based on the subgraph matching template, so as to obtain one or more target hyperedges conforming to the subgraph matching template; wherein the hypergraph structure comprises a plurality of hyperedges, each hyperedge connects a plurality of entity nodes to represent a composite event; determining a target paragraph list associated with the target hyperedge; constructing a target input sequence based on a target paragraph pointed by the target paragraph list and the query sentence, inputting the target input sequence into a natural language model, and generating a target answer text matched with the query sentence. The application can improve the retrieval accuracy and reliability, and also takes into account the retrieval efficiency.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Disease science popularization error correction method and system based on artificial intelligence

    CN120108774A

  • Intelligent search engine system and method based on NLP and vector hybrid retrieval

    CN121029791A

  • Knowledge base retrieval method fused with natural language large model

    CN121168677A