Intelligent matching method, device and equipment for man-on-duty process, and storage medium
By using open source large language model and human job process data, dynamic parameters are extracted and combined, and they are trained, an intelligent human job matching model is obtained, which solves the problems of inefficient and insufficient accuracy of human job matching in the existing technology, and achieves more efficient and accurate human job matching.
Patent Information
- Application Number
- CN202411701728.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-26
- Publication Date
- 2025-03-18
AI Technical Summary
The existing technology is inefficient in person-position matching, and the traditional employee scoring method lacks a prediction model with high accuracy, making it difficult to achieve accurate person-position matching.
Using an open source large language model, by obtaining human-job process data and converting it into a computable data matrix, dynamic parameters such as employee evaluation results, human-computer interface template serial number, repeated keywords and human-job matching subscene number are extracted, and a new data result set is combined to train the large language model to obtain an intelligent human-job matching model, which is used to calculate the human-job matching result.
It improves the accuracy and efficiency of human-job matching, reduces the allocation of computing resources by large models, and overcomes the problem that large model training fails to obtain accurate matching results many times.
Smart Images
Figure CN119647852A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of intelligent matching of people and positions, and in particular to a method, device, equipment and storage medium for intelligent matching of people and positions and processes. Background Art
[0002] With the surge in business volume and the large number of employees, the company still uses manual assignment to match employees to the most suitable positions, which is very inefficient. Although some teams have developed some systems to improve efficiency, employee rating indicators are also very important to enterprises and are an important basis for enterprises to optimize employees and match people with positions. However, traditional employee ratings are simply a simple weighted sum of employees' past performance. Although some teams have optimized the simple weighted sum, these solutions are only for simple intelligent matching and management of employee positions and lack a high-accuracy prediction model. Summary of the invention
[0003] Purpose of the invention: To propose a method, device, equipment and storage medium for intelligent matching of people, positions and processes to solve the above-mentioned problems existing in the prior art.
[0004] First, a method for intelligent matching of people, positions and processes is proposed, with the following steps:
[0005] Select an open source large language model LLM;
[0006] Acquire personnel job process data, and convert the personnel job process data into a computable data matrix;
[0007] Extracting four dynamic parameters, namely, employee evaluation results of each circulation and approval link, human-machine interface template serial number, repeated keywords and person-job matching sub-scenario number, from the data matrix, and combining the four dynamic parameters to form a new data result set;
[0008] Using the new data result set to train the large language model LLM to obtain an intelligent person-job matching model;
[0009] The intelligent job-person matching model is used to calculate and obtain the job-person matching result.
[0010] In a further embodiment, after the calculation of the person-job matching result is completed, if an abnormality occurs in the distributed link and a rollback occurs, the two-step spelling correction model deployed in the intelligent person-job matching model is activated to correct the content in the rollback link.
[0011] In a further embodiment, the personnel process data is converted into a computable data matrix, and the data matrix is in the following form:
[0012]
[0013] Where P m,m Represents the mth group of data under the mth layer.
[0014] In a further embodiment, an employee feature vector A=(a1, a2, ..., an) is extracted from the data matrix;
[0015] Four dynamic parameters, namely, employee evaluation results of each circulation and approval link, human-machine interface template serial number, repeated keywords and person-job matching sub-scenario number, are extracted from the data matrix. The four dynamic parameters are combined to form a new data result set. The new data result set is in the following form:
[0016] B={b1,b2,b3,b4}
[0017] In the formula, b1 represents the employee evaluation result, b2 represents the human-machine interface template serial number, b3 represents the repeated keywords, and b4 represents the person-job matching sub-scenario number.
[0018] In a further embodiment, the matching score between the employee feature vector and the new data result set is calculated by the cosine similarity formula:
[0019]
[0020] In the formula, the value range of CS is [-1,1]. The closer the CS value is to 1, the more similar the two vectors are.
[0021] In a further embodiment, the large language model LLM is trained using the new data result set to obtain an intelligent person-job matching model, and the L1 regression loss function of the intelligent person-job matching model is as follows:
[0022]
[0023] In the formula, Represents the expected value of a given input x, random noise z, and a preset reference true value y.
[0024] In a further embodiment, a conditional adversarial loss function is introduced Its expression is as follows:
[0025]
[0026] In the formula, represents the expected value of the given input x, random noise z and the preset reference truth value y; G represents trying to minimize the loss; D represents trying to maximize the loss; z represents random noise; represents the logarithmic probability of the expected output of the discriminator for the preset reference truth value y; It represents the log probability of the expected output of the discriminator for the data G(x,z) generated by the generator.
[0027] In a further embodiment, the conditional adversarial loss function and the L1 regression loss function Get the final loss function G * :
[0028]
[0029] In the formula, λ L1 represents the weight of L1 loss; argmin G max D It means that both the generation loss is minimized and the discrimination loss is maximized.
[0030] In a second aspect of the present invention, a device for intelligently matching people, positions and processes is provided, the device comprising:
[0031] Model access module, used to access the open source large language model LLM;
[0032] A data acquisition module, used to acquire personnel and job process data and convert the personnel and job process data into a computable data matrix;
[0033] A data analysis module is used to extract four dynamic parameters, namely, employee evaluation results of each circulation and approval link, human-machine interface template serial number, repeated keywords and person-job matching sub-scenario number, from the data matrix, and the four dynamic parameters are combined to form a new data result set;
[0034] A model training module, used to train the large language model LLM using the new data result set to obtain an intelligent person-job matching model;
[0035] The output module is used to access the intelligent job matching model to obtain the job matching result.
[0036] According to a third aspect of the present invention, an electronic device is provided, comprising: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, the method for intelligent matching of people, positions and processes as described in the first aspect is implemented.
[0037] According to a fourth aspect of the present invention, a computer-readable storage medium is provided, in which at least one executable instruction is stored. When the executable instruction is executed on an electronic device, the electronic device executes the intelligent matching method for people and job processes as described in the first aspect.
[0038] Beneficial effects: The present invention innovatively constructs an intelligent job matching module based on a large model, extracts four dynamic parameters, namely employee evaluations in each circulation and approval link, human-machine interface template serial numbers, repeated keywords, and job matching sub-scenario, from the first training result data matrix of the large model, and puts them into a small model for calculation to obtain a more intelligent job matching, thereby distributing some repeated and frequent processes to the automated job matching process to improve the timeliness of process processing. It overcomes the problem that the accurate matching result matrix cannot be obtained after multiple trainings of the large model. It reduces the threshold for the large model to configure high computing resources. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] Figure 1 It is a flow chart of the method for intelligent matching of people, positions and processes in an embodiment of the present invention.
[0040] Figure 2 It is a structural diagram of the intelligent matching device for personnel and job processes in the embodiment.
[0041] Figure 3 Schematic diagram of the structure of the electronic device in the embodiment. DETAILED DESCRIPTION
[0042] In the following description, a large number of specific details are provided to provide a more thorough understanding of the present invention. However, it is apparent to those skilled in the art that the present invention can be implemented without one or more of these details. In other examples, in order to avoid confusion with the present invention, some technical features known in the art are not described.
[0043] Before describing the embodiments, some terms that will appear later are first explained.
[0044] Large Language Models (LLMs): refers to deep learning models with a large number of parameters in the field of artificial intelligence, especially natural language processing (NLP). These models can understand and generate human language by learning massive text data from the Internet, showing a high level of language understanding and generation capabilities.
[0045] Person-job matching: refers to the process of optimizing the matching of individual characteristics such as personal ability, personality, values, interests, work experience, etc. with the job requirements, responsibilities, environment, culture, etc. in human resource management. This process aims to ensure that the most suitable employee can be found for each position, and at the same time, each employee can maximize his or her potential in his or her position, thus achieving the ideal state of "people are assigned the right talents and positions are assigned the right people".
[0046] See Figure 1 This embodiment discloses a method for intelligent matching of personnel, positions and processes, and the steps are as follows:
[0047] Firstly, an open source large language model (LLM) is selected; then, the person-job process data is obtained and converted into a computable data matrix; four dynamic parameters, including employee evaluation results of each flow and approval link, human-machine interface template serial number, repeated keywords and person-job matching sub-scenario number, are extracted from the data matrix, and the four dynamic parameters are combined to form a new data result set; then, the large language model (LLM) is trained using the new data result set to obtain an intelligent person-job matching model; finally, the intelligent person-job matching model is used to calculate and obtain the person-job matching result.
[0048] As a preferred embodiment, after the calculation of the person-job matching result is completed, if an abnormality occurs in the distributed link and a rollback occurs, the two-step spelling correction model deployed in the intelligent person-job matching model is started to correct the content in the rollback link.
[0049] As a preferred embodiment, the personnel process data is converted into a computable data matrix, and the data matrix is in the following form:
[0050]
[0051] Where P m,m Represents the mth group of data under the mth layer.
[0052] As a preferred embodiment, an employee feature vector A=(a1, a2, ..., an) is extracted from the data matrix;
[0053] Four dynamic parameters, namely, employee evaluation results of each circulation and approval link, human-machine interface template serial number, repeated keywords, and person-job matching sub-scenario number, are extracted from the data matrix. The four dynamic parameters are combined to form a new data result set. The new data result set is in the following form:
[0054] B={b1,b2,b3,b4}
[0055] In the formula, b1 represents the employee evaluation result, b2 represents the human-machine interface template serial number, b3 represents the repeated keywords, and b4 represents the person-job matching sub-scenario number.
[0056] As a preferred embodiment, the matching score between the employee feature vector and the new data result set is calculated by the cosine similarity formula:
[0057]
[0058] In the formula, the value range of CS is [-1,1]. The closer the CS value is to 1, the more similar the two vectors are.
[0059] As a preferred embodiment, the large language model LLM is trained using the new data result set to obtain an intelligent person-job matching model. The L1 regression loss function of the intelligent person-job matching model is as follows:
[0060]
[0061] In the formula, Represents the expected value of a given input x, random noise z, and a preset reference true value y.
[0062] As a preferred embodiment, the conditional adversarial loss function is introduced Its expression is as follows:
[0063]
[0064] In the formula, represents the expected value of the given input x, random noise z and the preset reference truth value y; G represents trying to minimize the loss; D represents trying to maximize the loss; z represents random noise; represents the logarithmic probability of the expected output of the discriminator for the preset reference truth value y; It represents the log probability of the expected output of the discriminator for the data G(x,z) generated by the generator.
[0065] As a preferred embodiment, the conditional adversarial loss function is combined and L1 regression loss function Get the final loss function G * :
[0066]
[0067] In the formula, λ L1 represents the weight of L1 loss; argmin G max D It means that both the generation loss is minimized and the discrimination loss is maximized.
[0068] Figure 2 What is shown is the intelligent matching device 400 of people and job processes proposed in this embodiment, which includes a model access module 401, a data acquisition module 402, a data analysis module 403, a model training module 404, and an output module 405.
[0069] The model access module 401 is used to access the open source large language model LLM; the data acquisition module 402 is used to obtain the human-job process data and convert the human-job process data into a computable data matrix; the data analysis module 403 is used to extract four dynamic parameters, namely, employee evaluation results of each flow approval link, human-computer interface template serial number, repeated keywords and human-job matching sub-scenario number, from the data matrix, and the four dynamic parameters are combined to form a new data result set; the model training module 404 is used to train the large language model LLM using the new data result set to obtain an intelligent human-job matching model; the output module 405 is used to access the intelligent human-job matching model to calculate and obtain the human-job matching result.
[0070] Figure 3 The internal structure diagram of the electronic device. The electronic device includes a processor, a memory, a communication interface and a communication bus. The processor, the memory and the communication interface communicate with each other through the communication bus. The memory is used to store at least one executable instruction, which enables the processor to execute Figure 1 The operating steps of the intelligent matching method for middle-level job processes.
[0071] The electronic device may also communicate with one or more external devices (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may communicate with one or more devices that enable a user to interact with the electronic device, and / or may communicate with any device that enables the electronic device to communicate with one or more other computing devices (e.g., routers, modems, etc.). This communication may be performed through an input / output (I / O) interface. In addition, the electronic device may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) through a network adapter. The network adapter communicates with other modules of the electronic device through a bus. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.
[0072] In practical applications, the operation logic of the method for intelligent matching of human-job processes disclosed in the above embodiment can be written into a set of executable instructions, which can be written into a storage medium and run on an electronic device. More specific examples of the computer-readable storage medium mentioned in this embodiment may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. The computer-readable storage medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a readable program code is carried. This propagated data signal can take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The readable signal medium can also be any readable medium other than a readable storage medium, which can send, propagate, or transmit a program for use by an instruction execution system, an apparatus, or a device or used in combination with it.
[0073] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above embodiments are not described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification. Although the present invention has been represented and described with reference to specific preferred embodiments, it shall not be interpreted as a limitation of the present invention itself. Various changes can be made to it in form and detail without departing from the spirit and scope of the present invention defined in the attached claims.
Claims
1. A method for intelligent matching of people, positions and processes, characterized in that: The steps include: Select an open source large language model LLM; Acquire personnel job process data, and convert the personnel job process data into a computable data matrix; Extracting four dynamic parameters, namely, employee evaluation results of each circulation and approval link, human-machine interface template serial number, repeated keywords and person-job matching sub-scenario number, from the data matrix, and combining the four dynamic parameters to form a new data result set; Using the new data result set to train the large language model LLM to obtain an intelligent person-job matching model; The intelligent job-person matching model is used to calculate and obtain the job-person matching result.
2. According to claim 1, a method for intelligent matching of personnel and job processes is characterized in that: After the calculation of the person-job matching results is completed, if an exception occurs in the distributed link and a rollback occurs, the two-step spelling correction model deployed in the intelligent person-job matching model is started to correct the content in the rollback link.
3. The method for intelligent matching of personnel and job processes according to claim 1 is characterized in that: The personnel process data is converted into a computable data matrix, the data matrix is in the following form: Where P m,m Represents the mth group of data under the mth layer.
4. The method for intelligent matching of personnel and job processes according to claim 3 is characterized in that: Extracting employee feature vector A=(a1, a2, ..., an) from the data matrix; Four dynamic parameters, namely, employee evaluation results of each circulation and approval link, human-machine interface template serial number, repeated keywords and person-job matching sub-scenario number, are extracted from the data matrix. The four dynamic parameters are combined to form a new data result set. The new data result set is in the following form: B={b1,b2,b3,b4} In the formula, b1 represents the employee evaluation result, b2 represents the human-machine interface template serial number, b3 represents the repeated keywords, and b4 represents the person-job matching sub-scenario number.
5. The method for intelligent matching of personnel and job processes according to claim 4 is characterized in that: The cosine similarity formula is used to calculate the matching score between the employee feature vector and the new data result set: In the formula, the value range of CS is [-1,1]. The closer the CS value is to 1, the more similar the two vectors are.
6. The method for intelligent matching of personnel and job processes according to claim 1 is characterized in that: The large language model LLM is trained using the new data result set to obtain an intelligent person-job matching model. The L1 regression loss function of the intelligent person-job matching model is as follows: In the formula, Represents the expected value of a given input x, random noise z, and a preset reference true value y.
7. The method for intelligent matching of personnel and job processes according to claim 6 is characterized in that: Introducing conditional adversarial loss function Its expression is as follows: In the formula, represents the expected value of the given input x, random noise z and the preset reference truth value y; G represents trying to minimize the loss; D represents trying to maximize the loss; z represents random noise; represents the logarithmic probability of the expected output of the discriminator for the preset reference truth value y; represents the logarithmic probability of the expected output of the discriminator for the data G(x,z) generated by the generator; Combine the above conditions to fight against the loss function and the L1 regression loss function Get the final loss function G * : In the formula, λ L1 represents the weight of L1 loss; argmin G max D It means that both the generation loss is minimized and the discrimination loss is maximized.
8. An intelligent matching device for personnel, job and process, characterized in that: include: Model access module, used to access the open source large language model LLM; A data acquisition module, used to acquire personnel and job process data and convert the personnel and job process data into a computable data matrix; A data analysis module is used to extract four dynamic parameters, namely, employee evaluation results of each circulation and approval link, human-machine interface template serial number, repeated keywords and person-job matching sub-scenario number, from the data matrix, and the four dynamic parameters are combined to form a new data result set; A model training module, used to train the large language model LLM using the new data result set to obtain an intelligent person-job matching model; The output module is used to access the intelligent job matching model to obtain the job matching result.
9. An electronic device, characterized in that: The device comprises: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, the method for intelligent matching of people and job processes as described in any one of claims 1 to 7 is implemented.
10. A computer-readable storage medium, characterized in that: The storage medium stores at least one executable instruction. When the executable instruction is executed on the electronic device, the electronic device executes the person-job process intelligent matching method as described in any one of claims 1 to 7.