A method for controlling an inference phase of multi-element customized text generation

By employing an adaptive guidance strategy of factor decomposition and dynamic scoring, the problems of unbalanced factor management and semantic deviation in multi-factor customized text generation are solved, thus achieving stability and consistency in text generation.

CN122366416APending Publication Date: 2026-07-10XIDIAN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
XIDIAN UNIV
Filing Date
2026-03-30
Publication Date
2026-07-10

AI Technical Summary

Technical Problem

Existing technologies struggle to achieve precise and dynamic process control for various customized needs in long text or multi-round generation scenarios, leading to problems such as customized feature decay, uneven response, semantic deviation, and inconsistent generation.

Method used

By using factor decomposition and low-redundancy validation, multiple customized elements are decomposed into independent factors, which are then dynamically scored and fused to construct an adaptive guidance strategy. In the iterative generation process, factor weights are quantitatively evaluated to achieve stable control over the probability distribution.

Benefits of technology

It enables independent management of elements, adaptive stage adjustment, and suppression of redundant interference in multi-element customized text generation, thereby improving the stability and consistency of generation quality.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122366416A_ABST
    Figure CN122366416A_ABST
Patent Text Reader

Abstract

This invention provides a method for controlling the inference stage of multi-factor customized text generation, comprising: dynamically scoring and fusion calculating each factor based on the currently generated text and context factor verification information to obtain a fusion score; constructing an activation factor set based on the fusion score; normalizing the fusion scores corresponding to the factors in the activation factor set using a softmax function to obtain a final weight set; obtaining the probability distribution of the current position using the currently generated text, the final weight set, and a preset large language model; obtaining the current output word based on the probability distribution of the current position, and adding the current output word to the currently generated text as the current generated text in the above process; and repeating the above process until the final generated text is generated. This method achieves independent management of elements, adaptive stage adjustment, redundancy interference suppression, and stable optimization of effect and quality in multi-factor customized text generation.
Need to check novelty before this filing date? Find Prior Art