A Chinese text correction method based on model fusion and scene self-adaption

By using parallel encoding of cross-model attention and Chinese character structure awareness text scene recognition algorithms, and integrating BERT, ERNIE and T5 models, a three-path mutual error correction model is constructed and character-by-character voting is performed. This solves the problems of insufficient language representation and scene adaptation in existing Chinese error correction models, and achieves efficient self-adaptive error correction for specific scenes.

CN120975078BActive Publication Date: 2026-03-03CHINA ORDINS GRP CO LTD
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202511090344.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-08-05
Publication Date
2026-03-03
Estimated Expiration
2045-08-05

AI Technical Summary

Technical Problem

Existing Chinese error correction models suffer from limited language representation capabilities and insufficient generalization ability when used individually. Traditional model cascading methods suffer from information redundancy and static fusion issues. Furthermore, general error correction models are not optimized for specific scenarios, resulting in limited error correction performance.

Method used

We adopt a model fusion and scene self-adaptation approach, which integrates BERT, ERNIE and T5 models through parallel encoding of cross-model attention paths, and combines them with the Chinese character structure perception text scene recognition algorithm CSA-TextCNN to construct a three-path mutual error correction model. We then perform character-by-character voting, and load the corresponding parameters for error correction after recognizing the text scene.

Benefits of technology

It improves language representation capabilities, mitigates the error correction bias and data noise effects of a single model, achieves self-adaptive error correction performance optimization in specific scenarios, solves information redundancy and static fusion problems, and improves error correction accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120975078B_ABST
    Figure CN120975078B_ABST
Patent Text Reader

Abstract

The present application relates to a kind of Chinese text error correction method based on model fusion and scene self-adaption, belong to electric digital data processing field, solve the limited language representation ability of single model and the insufficient generalization ability, traditional model cascade mode exists information redundancy and static fusion, while general error correction model is not optimized for specific scene, leading to error correction performance is limited.The problem of including the input of text to be corrected text training good text scene identification model, obtains corresponding scene category;Based on scene category, load the corresponding training good three-path mutual error correction model parameters of this scene;The text to be corrected text is input into the three-path mutual error correction model after loading parameters, to generate first, second and third error correction results;Wherein, three-path mutual error correction model includes parallel first, second and third mutual error correction path;Based on first, second and third error correction results, word-by-word voting is carried out, and the text after error correction corresponding to the text to be corrected text is obtained.The realization of high-efficiency error correction of scene self-adaption.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Text error correction model training method, application method, device and equipment

    CN115146621A

  • Text error correction method and device based on GRU-Transform, electronic equipment and storage medium

    CN120106053A