Learning course recommendation method based on age of worker
By using a deep learning-based course recommendation method based on worker age, the problem of failing to consider age differences in existing technologies has been solved, enabling accurate course recommendations and improving worker learning efficiency and enterprise resource utilization.
Patent Information
- Application Number
- CN202510903607.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-01
- Publication Date
- 2025-11-25
AI Technical Summary
Existing course recommendation methods fail to adequately consider the age differences among workers, resulting in a mismatch between recommendations and actual needs, leading to a waste of learning resources and a decline in workers' learning enthusiasm.
By collecting data such as workers' age and occupation, and using the entropy weight method to calculate feature weights, a deep learning-based course recommendation model is constructed to make accurate course recommendations based on workers' age groups.
This significantly improved the accuracy of course recommendations and the efficiency of worker learning resource utilization, thereby enhancing worker learning enthusiasm and the effectiveness of enterprise resource allocation.
Smart Images
Figure CN121010068A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of smart community service technology, and more specifically, to a method for recommending learning courses based on workers' age. Background Technology
[0002] In today's era of rapid development in digital learning, enterprises and online learning platforms widely adopt course recommendation systems to improve user learning experience and resource utilization efficiency. Existing course recommendation technologies are mostly based on algorithms such as collaborative filtering, content-based recommendation, and hybrid recommendation. These algorithms primarily rely on users' historical behavioral data, such as course browsing history, learning duration, favorites, and ratings, as well as course text descriptions and tags. They make recommendations by analyzing the similarity between users or the matching degree between users and courses. For example, collaborative filtering algorithms find other users with similar learning behaviors to the target user and recommend courses liked by these similar users to the target user; content-based recommendation algorithms match user interests based on course content features, such as knowledge points and difficulty levels. However, these traditional recommendation algorithms have certain limitations in practical applications. They often ignore the differences among user groups and fail to fully consider the essential differences in learning needs, learning abilities, and learning styles among different types of users, resulting in insufficient targeting and accuracy of the recommendation results.
[0003] For the specific user group of workers, age has a significant and complex impact on their learning needs and abilities. As workers age, they exhibit different characteristics in terms of cognitive abilities, knowledge reserves, and career development stages. Young workers are typically energetic, have a strong ability to absorb new knowledge and technologies, and are eager to master cutting-edge skills to gain a competitive edge in the early stages of their careers. Middle-aged workers are in a stable period of career development and focus more on deepening their professional skills and improving their management abilities to meet the demands of job advancement. Older workers may face difficulties in updating their knowledge and declining physical functions, requiring more learning related to safe operation of their jobs and basic use of new equipment to ensure smooth work performance. However, existing course recommendation methods do not take workers' age into core consideration, failing to accurately identify the personalized learning needs of workers at different age stages. This results in a mismatch between recommended courses and workers' actual needs, wasting learning resources and dampening workers' learning enthusiasm, hindering effective support for workers' professional skill improvement and corporate talent development strategies.
[0004] Therefore, there is an urgent need for a course recommendation method that can fully take into account the age differences of workers, in order to fill the gap in existing technology in this field.
[0005] No effective solutions have yet been proposed to address the problems in the relevant technologies. Summary of the Invention
[0006] In response to the problems in related technologies, this invention proposes a learning course recommendation method based on worker age to overcome the aforementioned technical problems existing in the current related technologies.
[0007] The technical solution of this invention is implemented as follows:
[0008] A method for recommending learning courses based on worker age includes the following steps:
[0009] Collect workers' age, occupation, years of service, learning history, job skill requirements, learning goals and interests, and quantify the data to form feature vectors;
[0010] Workers were divided into youth, middle-aged, and elderly groups based on their age.
[0011] Based on workers of different age groups, features such as occupational information, years of service, learning history, job skill requirements, learning goals and interest preferences are extracted, and the entropy weight method is used to calculate the weight of each feature.
[0012] A course recommendation model is constructed, which includes an input layer, a hidden layer, and an output layer. The hidden layer uses the ReLU activation function, and the output layer outputs the recommendation score for each course.
[0013] The worker's age group is determined based on their age. Other characteristic data of the worker are then input into the trained course recommendation model to calculate courses with a high degree of matching with the worker. A course recommendation list is then generated in descending order of matching degree.
[0014] Furthermore, the learning history includes courses taken, learning duration, course completion rate, and learning grades.
[0015] Furthermore, the job skill requirements information is determined through enterprise job analysis and industry standards, while the interest and preference information is obtained through user-defined settings.
[0016] Furthermore, the age grouping function is expressed as:
[0017]
[0018] Furthermore, the calculation of the weights of each feature includes the following steps:
[0019] Let the feature value of the i-th sample under the j-th feature be x. ij After standardization, it becomes y ij The number of samples is m, and the number of features is n. The entropy value e of the j-th feature. j , represented as:
[0020]
[0021] in, The weight w of the j-th feature j , represented as:
[0022] Furthermore, the hidden layer of the course recommendation model is represented as follows:
[0023] H=σ(W xh X+b h );
[0024] Where σ is the activation function, a ReLU function, expressed as: σ(x)=max(0,x), W xh Let b be the hidden layer weight matrix. h This is the hidden layer bias vector.
[0025] Furthermore, the output layer is represented as:
[0026] Y = W hy H+b y ;
[0027] Among them, W hy Let b be the output layer weight matrix. y This is the output layer bias vector.
[0028] Furthermore, the method includes the following steps: training the course recommendation model using historical data, employing a cross-entropy loss function to measure the difference between predicted and true values, and updating the model parameters using a backpropagation algorithm. The cross-entropy loss function L, used to measure the difference between predicted and true values, is expressed as:
[0029]
[0030] Where m is the sample size and K is the number of courses. The labels are real labels, indicating whether workers have taken courses k and y. ik The recommended score for course k predicted by the model.
[0031] Furthermore, generating the course recommendation list from high to low matching degree includes: determining the matching degree based on the output score of the course recommendation model, and generating the course recommendation list from high to low matching degree.
[0032] The beneficial effects of this invention are:
[0033] 1. This invention, through age grouping and feature weight calculation, can accurately pinpoint the personalized needs of workers at different age stages. By incorporating worker age as a core variable into the course recommendation system and combining multi-dimensional data and mathematical calculation models, the accuracy of course recommendations is significantly improved. For example, when young workers have a strong need to learn new technologies, the model can prioritize recommending courses on emerging technologies such as automated production and programming based on their learning interests and preferences. For older workers, the model provides targeted recommendations for safe operation and basic equipment usage courses based on their safety operating procedures and equipment operation proficiency. This precise matching greatly improves the alignment between courses and workers' needs, enabling workers to quickly access learning resources that are truly valuable for their career development. This effectively solves the problem of traditional recommendation methods that "cast a wide net" but fail to hit the right needs.
[0034] 2. This invention not only improves workers' learning efficiency but also brings significant benefits of resource optimization and cost reduction to enterprises and online learning platforms. Because the recommended courses are highly tailored to workers' needs, their learning enthusiasm and course completion rates are greatly enhanced, preventing a large amount of learning resources from being wasted due to mismatches. Taking manufacturing enterprises as an example, after workers learn courses that are suitable for their age and job requirements, their skills improve more rapidly, enabling them to apply their knowledge to work practice more quickly and create higher value for the enterprise. At the same time, for online learning platforms, accurate course recommendations reduce user churn caused by inaccurate recommendations, optimize platform resource allocation, reduce operating costs, and achieve a win-win situation for worker personal development, enterprise talent cultivation, and platform resource utilization, powerfully promoting the digital transformation and high-quality development of vocational education and corporate training. Attached Figure Description
[0035] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0036] Figure 1 This is a flowchart illustrating a learning course recommendation method based on worker age according to an embodiment of the present invention. Detailed Implementation
[0037] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention are within the scope of protection of the present invention.
[0038] According to an embodiment of the present invention, a method for recommending learning courses based on the worker's age is provided.
[0039] like Figure 1 As shown, the learning course recommendation method based on worker age according to an embodiment of the present invention includes the following steps:
[0040] Information such as workers' age, occupation, years of service, learning history, job skill requirements, and their own learning goals and interests is collected in advance. This data is then quantified to form a feature vector, represented as follows:
[0041] X = [x1, x2, ..., x n ];
[0042] Among them, X i Let i represent the i-th eigenvalue.
[0043] This technical solution includes a learning history record, such as courses taken, learning duration, course completion rate, and learning grades; job skill requirements information can be determined through enterprise job analysis and industry standards; and interest preference information can be obtained through user-defined settings.
[0044] Workers are divided into different age groups based on their age, including: youth group (18-30 years old), middle-aged group (31-50 years old), and elderly group (51 years old and above), represented as follows: Let age be a, the age grouping function is expressed as:
[0045]
[0046] Based on workers of different age groups, features such as occupational information, years of service, learning history, job skill requirements, learning goals, and interest preferences are extracted. Weights for each feature are calculated, including: setting the feature value of the i-th sample under the j-th feature as x. ij After standardization, it becomes y ij The number of samples is m, and the number of features is n. The entropy value e of the j-th feature. j , represented as:
[0047]
[0048] in, The weight w of the j-th feature j , represented as:
[0049] The course recommendation model is constructed, including: labeling the input layer as X, the hidden layer as H, the output layer as Y, the weight matrix as W, and the bias vector as b;
[0050] The hidden layer is represented as follows:
[0051] H=σ(W xh X+b h );
[0052] Where σ is the activation function, which is the ReLU function, expressed as: σ(x)=max(0,x);
[0053] The output layer is represented as:
[0054] Y = W hy H+b y ;
[0055] In Y, the elements represent the recommended scores for each course, with higher scores indicating a higher degree of match with the worker.
[0056] In addition, the model is trained using collected historical data, and the cross-entropy loss function L is used to measure the difference between the predicted and actual values, expressed as:
[0057]
[0058] Where m is the sample size and K is the number of courses. The labels are real labels, indicating whether workers have taken courses k and y. ik The recommended score for course k predicted by the model is used. The weight matrix W and bias vector b are updated through backpropagation to minimize the loss function L, thereby optimizing the model performance and enabling it to accurately predict the course preferences and needs of workers in different age groups.
[0059] The course recommendation process includes: determining the age group of the worker based on their age information, inputting other characteristic data of the worker into the constructed course recommendation model, calculating courses with a high degree of matching with the worker, and generating a course recommendation list in descending order of matching degree, which is then recommended to the worker.
[0060] Furthermore, in specific implementation, taking a manufacturing company as an example, the details are as follows:
[0061] A manufacturing company collected data on 500 workers through its internal management system and online learning platform. Age information was obtained from employee files; occupational information covered multiple positions including machining, equipment maintenance, and quality inspection; years of service ranged from 1 to 30 years; learning history included detailed information on various courses taken by workers in the past year; job skill requirements were determined based on the company's production processes and job requirements; and learning goals and interests were collected through company-organized questionnaires. After quantifying the data, a feature vector X was obtained for each worker. i , i = 1, 2, ..., 500.
[0062] The workers were divided into three groups: youth (18-30 years old, 150 people), middle-aged (31-50 years old, 250 people), and elderly (51 years old and above, 100 people).
[0063] Feature extraction was performed, including: for the young worker group, it was found that most workers showed a strong interest in emerging technology courses such as automated production technology and programming, and their learning progress was relatively fast. The entropy weight method was used to calculate feature weights, assuming that features such as learning interest preferences, learning history, and job skill requirements were calculated, resulting in weights w respectively. 兴趣 =0.4, w 历史 =0.3, w 需求 =0.2;
[0064] Among middle-aged workers, those in management positions showed a higher level of interest in management-related courses such as project management and team building. In terms of professional skills depth, workers proficient in operating specific equipment and improving processes demonstrated outstanding performance. The weight of the management-related course learning characteristic was calculated to be w. 管理 =0.35, the weight of the professional skills depth feature is w 深度 =0.3, the weight of the job skill requirement feature is w 需求 =0.2;
[0065] Among the older workers, the learning records of the safe operation procedure course were relatively concentrated, but some workers had difficulty learning the new equipment operation course. The feature weight of the safe operation procedure was determined to be w. 安全 =0.4, the weight of the equipment operation course mastery feature is w 设备 =0.3, the weight of the job skill requirement feature is w 需求 =0.2.
[0066] A deep learning neural network algorithm was employed, using TensorFlow as the framework, to construct a neural network with three hidden layers. Extracted features and weights of workers in each age group were used as input layer parameters. The number of neurons in the hidden layer was 128, 64, and 32, respectively, and the number of neurons in the output layer was the same as the number of courses. The model was trained using collected historical data with a learning rate of 0.001 and 100 iterations. Backpropagation was used to adjust the model parameters and optimize its performance, enabling the model to accurately predict the course preferences and needs of workers in different age groups.
[0067] Taking a 25-year-old young machining worker as an example, based on his age, he belongs to the youth group. His occupational information (machining), years of work experience (3 years), learning history (studied basic mechanical drawing for 10 hours, completion rate 80%, good grade), job skill requirements (mastering CNC machining technology), and learning goals and interests (hoping to learn automated machining technology) are input into a pre-constructed course recommendation model. The model calculates and outputs recommendation scores for each course, generating a course recommendation list from high to low scores. Top-ranked courses include "Advanced CNC Programming Course" and "Automated Production Line Operation and Maintenance Course," etc., and this recommendation list is pushed to the worker. After reviewing the recommended courses, the worker selects to study the "Automated Production Line Operation and Maintenance Course" and gives the course a positive review during the learning process.
[0068] In addition, taking another service industry company as an example, the same steps were followed for data collection, age grouping, feature extraction and analysis, construction of a course recommendation model, and course recommendations. Considering the characteristics of this industry, the feature extraction and analysis focused on features related to customer service skills and communication ability improvement. For example, younger service personnel were more inclined to learn emerging service skills courses such as new media marketing and online customer service; middle-aged service managers had a greater need for courses on team management and service process optimization; and older service personnel needed to learn basic service standards and customer complaint handling skills. The entropy weight method was used to calculate the weights of each feature, and a recommendation model was constructed using a deep learning neural network. Through practical application, accurate course recommendations were provided for service personnel of different age groups, effectively improving their professional skills and service quality, and receiving unanimous recognition from the company and its employees.
[0069] In summary, by means of the above-described technical solution of the present invention, through
[0070] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Those skilled in the art, upon considering the disclosure in the specification and embodiments, will readily conceive of other embodiments of this disclosure. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the claims.
[0071] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.
Claims
1. A method for recommending learning courses based on worker age, characterized in that, Includes the following steps: Collect workers' age, occupation, years of service, learning history, job skill requirements, learning goals and interests, and quantify the data to form feature vectors; Workers were divided into youth, middle-aged, and elderly groups based on their age. Based on workers of different age groups, features such as occupational information, years of service, learning history, job skill requirements, learning goals and interest preferences are extracted, and the entropy weight method is used to calculate the weight of each feature. A course recommendation model is constructed, which includes an input layer, a hidden layer, and an output layer. The hidden layer uses the ReLU activation function, and the output layer outputs the recommendation score for each course. The worker's age group is determined based on their age. Other characteristic data of the worker are then input into the trained course recommendation model to calculate courses with a high degree of matching with the worker. A course recommendation list is then generated in descending order of matching degree.
2. The learning course recommendation method based on worker age according to claim 1, characterized in that, The learning history includes courses taken, learning duration, course completion rate, and learning grades.
3. The learning course recommendation method based on worker age according to claim 2, characterized in that, The job skill requirements information is determined through enterprise job analysis and industry standards, while the interest preference information is obtained through user-defined settings.
4. The learning course recommendation method based on worker age according to claim 1, characterized in that, The age grouping function is expressed as:
5. The learning course recommendation method based on worker age according to claim 1, characterized in that, The calculation of the weights of each feature includes the following steps: Let the feature value of the i-th sample under the j-th feature be x. ij After standardization, it becomes y ij The number of samples is m, and the number of features is n. The entropy value e of the j-th feature. j , represented as: in, The weight w of the j-th feature j , represented as:
6. The learning course recommendation method based on worker age according to claim 1, characterized in that, The hidden layer of the course recommendation model is represented as follows: H=σ(W xh X+b h ); Where σ is the activation function, a ReLU function, expressed as: σ(x)=max(0,x), W xh Let b be the hidden layer weight matrix. h This is the hidden layer bias vector.
7. The learning course recommendation method based on worker age according to claim 6, characterized in that, The output layer is represented as: Y=W hy H+b y ; Among them, W hy Let b be the output layer weight matrix. y This is the output layer bias vector.
8. The learning course recommendation method based on worker age according to claim 1, characterized in that, It also includes the following steps: The course recommendation model is trained using historical data, and the cross-entropy loss function is used to measure the difference between predicted and true values. The model parameters are then updated using a backpropagation algorithm. The cross-entropy loss function L, used to measure the difference between predicted and true values, is expressed as: Where m is the sample size and K is the number of courses. The labels are real labels, indicating whether workers have taken courses k and y. ik The recommended score for course k predicted by the model.
9. The learning course recommendation method based on worker age according to claim 1, characterized in that, The process of generating a course recommendation list from high to low matching degree includes: determining the matching degree based on the output score of the course recommendation model, and generating a course recommendation list from high to low matching degree.