CN-122022747-A - Human resource intelligent training method and system based on LLM

Abstract

The invention provides a human resource intelligent training method and system based on LLM (logical link management), which are used for acquiring historical staff training information, carrying out preliminary processing and training a first LLM training model, acquiring enterprise training requirement information, outputting a training knowledge base according to the first LLM training model, training staff according to the training knowledge base to obtain a first training result, calculating a training matching rate according to the training result, further carrying out specific fine adjustment on the first LLM training model to obtain an enterprise regulation model, calculating staff knowledge conversion rate, obtaining staff knowledge short-board information, setting training vectors according to the staff knowledge short-board information, and carrying out secondary specific fine adjustment on the first LLM training model through the training vectors to obtain a staff regulation model.

Inventors

• XU HUI
• LI XIANGYU
• Guo Jieling

Assignees

• 华南理工大学

Dates

Publication Date

20260512

Application Date

20251204

Claims (10)

1. 1. An intelligent training method for human resources based on LLM, which is characterized by comprising the following steps: S1, acquiring historical staff training information, and primarily processing the historical staff training information to train a first LLM training model; S2, acquiring enterprise training requirement information, outputting a training knowledge base according to a first LLM training model, and training staff according to the training knowledge base to acquire a first training result; s3, calculating a training matching rate according to a training result, and analyzing the training matching rate to further perform specific fine adjustment on the first LLM training model to obtain an enterprise regulation model; S4, calculating the employee knowledge conversion rate, obtaining employee knowledge short-board information, setting training vectors according to the employee knowledge short-board information, and performing secondary specific fine adjustment on the first LLM training model through the training vectors to obtain an employee adjustment model.
2. 2. The LLM-based human resource intelligent training method as set forth in claim 1, wherein S1 comprises: S101, acquiring historical staff training information, preprocessing the historical staff training information, and classifying according to staff posts to obtain a plurality of sub-information groups; s102, training a first LLM training model through a plurality of sub-information groups, wherein the first LLM training model can output training knowledge bases of the plurality of sub-information groups.
3. 3. The LLM-based human resource intelligent training method as set forth in claim 1, wherein S2 comprises: S201, collecting enterprise training requirement information, preprocessing the enterprise training requirement information to obtain processing information, inputting the processing information into a first LLM training model, and obtaining a training knowledge base corresponding to the enterprise training requirement information; S202, generating enterprise demand knowledge questions through the training knowledge base to train staff, and obtaining staff training scores, namely a first training result.
4. 4. The LLM-based human resource intelligent training method as set forth in claim 1, wherein S3 comprises: S301, calculating a training matching rate according to a first training result of staff, comparing the training matching rate with a preset matching rate threshold, and judging whether a training knowledge base learned by the corresponding staff is matched with the enterprise training requirement information according to a comparison result to obtain a judgment result; S302, calling a preset strategy according to a judging result, and fine-tuning the first LLM training model through the preset strategy to obtain an enterprise regulation model.
5. 5. The LLM-based human resource intelligent training method as set forth in claim 1, wherein S4 comprises: S401, calculating employee knowledge conversion rate through employee training information, and acquiring employee knowledge short-board information through the employee knowledge conversion rate; s402, setting staff training vectors according to staff knowledge short-board information, fine-tuning the first LLM training model according to the staff training vectors to obtain a staff adjustment model, and outputting a staff exclusive knowledge base to staff through the staff adjustment model.
6. 6. An LLM-based human resources intelligent training system, the system comprising: The first training module is used for acquiring historical staff training information, and is used for training a first LLM training model by carrying out preliminary processing on the historical staff training information; The first training module is used for acquiring the enterprise training requirement information, outputting a training knowledge base according to a first LLM training model, and training staff according to the training knowledge base to acquire a first training result; the first adjustment module is used for calculating a training matching rate according to a training result, analyzing the training matching rate, and further carrying out specific fine adjustment on the first LLM training model to obtain an enterprise adjustment model; the second adjusting module is used for calculating the employee knowledge conversion rate, obtaining employee knowledge short-board information, setting training vectors according to the employee knowledge short-board information, and carrying out secondary specific fine adjustment on the first LLM training model through the training vectors to obtain an employee adjusting model.
7. 7. The LLM based human resource intelligent training system as set forth in claim 6, wherein the steps of: The information grouping module is used for acquiring historical staff training information, preprocessing the historical staff training information, classifying according to staff posts and acquiring a plurality of sub-information groups; The model training module is used for training a first LLM training model through a plurality of sub-information groups, and the first LLM training model can output a training knowledge base of the plurality of sub-information groups.
8. 8. The LLM based human resource intelligent training system as set forth in claim 6, wherein the steps of: The knowledge base acquisition module is used for collecting enterprise training requirement information, preprocessing the enterprise training requirement information to obtain processing information, inputting the processing information into the first LLM training model, and obtaining a training knowledge base corresponding to the enterprise training requirement information; And the training module is used for generating enterprise demand knowledge questions through the training knowledge base to train staff and obtaining staff training scores, namely a first training result.
9. 9. The LLM based human resource intelligent training system as set forth in claim 6, wherein the steps of: The judging module is used for calculating training matching rate according to the first training result of the staff, comparing the training matching rate with a preset matching rate threshold value, judging whether a training knowledge base learned by the corresponding staff is matched with the enterprise training requirement information according to the comparison result, and obtaining a judging result; And the enterprise regulation module is used for calling a preset strategy according to the judgment result, and carrying out fine adjustment on the first LLM training model through the preset strategy to obtain an enterprise regulation model.
10. 10. The LLM based human resource intelligent training system as set forth in claim 6, wherein the steps of: The conversion module is used for calculating employee knowledge conversion rate according to the employee training information and obtaining employee knowledge short-board information according to the employee knowledge conversion rate; And the staff adjustment module is used for setting staff training vectors according to staff knowledge short-board information, carrying out fine adjustment on the first LLM training model according to the staff training vectors to obtain a staff adjustment model, and outputting a staff exclusive knowledge base to staff through the staff adjustment model.

Description

Human resource intelligent training method and system based on LLM Technical Field The invention provides a human resource intelligent training method and system based on LLM, relates to the technical field of intelligent training, and particularly relates to the technical field of intelligent training of human resources based on LLM. Background With the rapid development of artificial intelligence technology, natural Language Processing (NLP) is increasingly used in various fields. Among them, a Large Language Model (LLM), which is an advanced NLP technology, has shown a strong capability in text generation, dialog systems, machine translation, and the like. In recent years, more and more researchers and enterprises have begun to explore how LLM can be applied to the field of staff training to improve the efficiency and quality of training. In traditional employee training, enterprises often develop training plans and content empirically and intuitively, lacking scientific and quantitative methods to assess the effectiveness of the training. In addition, individual differences among staff are often ignored, so that training contents are disjointed from actual demands, and methods of how to conduct targeted training according to the individual differences of the staff and knowledge shortboards, how to optimize and adjust models and the like are provided. Disclosure of Invention The invention provides a human resource intelligent training method and system based on LLM, which are used for solving the problems that in the traditional staff training, enterprises usually make training plans and contents according to experience and intuition, and a scientific and quantitative method is lacked to evaluate the training effect. In addition, individual differences among staff are often ignored, so that training contents are disjointed from actual demands, and the problems of methods and the like on how to conduct targeted training according to the individual differences of the staff and knowledge shortboards, how to optimize and adjust models and the like are solved: The invention provides a human resource intelligent training method and a system based on LLM, wherein the method comprises the following steps: S1, acquiring historical staff training information, and primarily processing the historical staff training information to train a first LLM training model; S2, acquiring enterprise training requirement information, outputting a training knowledge base according to a first LLM training model, and training staff according to the training knowledge base to acquire a first training result; s3, calculating a training matching rate according to a training result, and analyzing the training matching rate to further perform specific fine adjustment on the first LLM training model to obtain an enterprise regulation model; S4, calculating the employee knowledge conversion rate, obtaining employee knowledge short-board information, setting training vectors according to the employee knowledge short-board information, and performing secondary specific fine adjustment on the first LLM training model through the training vectors to obtain an employee adjustment model. Further, the step S1 includes: S101, acquiring historical staff training information, preprocessing the historical staff training information, and classifying according to staff posts to obtain a plurality of sub-information groups; s102, training a first LLM training model through a plurality of sub-information groups, wherein the first LLM training model can output training knowledge bases of the plurality of sub-information groups. Further, the step S2 includes: S201, collecting enterprise training requirement information, preprocessing the enterprise training requirement information to obtain processing information, inputting the processing information into a first LLM training model, and obtaining a training knowledge base corresponding to the enterprise training requirement information; S202, generating enterprise demand knowledge questions through the training knowledge base to train staff, and obtaining staff training scores, namely a first training result. Further, the step S3 includes: S301, calculating a training matching rate according to a first training result of staff, comparing the training matching rate with a preset matching rate threshold, and judging whether a training knowledge base learned by the corresponding staff is matched with the enterprise training requirement information according to a comparison result to obtain a judgment result; S302, calling a preset strategy according to a judging result, and fine-tuning the first LLM training model through the preset strategy to obtain an enterprise regulation model. Further, the step S4 includes: S401, calculating employee knowledge conversion rate through employee training information, and acquiring employee knowledge short-board information through the employee knowledge conversion rate; s