Online education learning path recommendation system for environmental art design major

Abstract

The invention provides an environment art design major online education learning path recommendation system, and relates to the technical field of online education, and the system comprises a user information obtaining module which is used for obtaining the basic information, learning background, interest preference and learning target of a user; the course resource database module is used for storing various online course resources of the environmental art design major, including course videos, teaching courseware, practice items and case analysis; and the learning behavior analysis module is used for analyzing behavior characteristics of the user in the learning process, including learning styles, learning habits and content preferences, and acquiring a learning behavior mode of the user. According to the method, the user information is comprehensively acquired, the learning behaviors are analyzed, the personalized learning path is determined, and real-time evaluation and adjustment are performed, so that the learning efficiency, satisfaction and achievement of the user are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of online education, and in particular to an online education learning path recommendation system for environmental art design major. Background Art

[0002] Traditional online education relies on fixed teaching resources, such as pre-recorded video courses and e-books. Although these resources are quite abundant in quantity, they are often updated slowly and sometimes fail to reflect the latest trends and technological developments in the industry in a timely manner. For example, in the major of environmental art design, students may not be able to learn about the new features of the latest design software or the latest trends in the design industry through these resources.

[0003] Secondly, traditional online education often adopts a "one-size-fits-all" teaching method, which is difficult to fully meet the learning needs and styles of different students. Because students have different learning styles and paces, some students may prefer to learn visually, while others prefer to master knowledge through practice. However, traditional online education often cannot provide such personalized teaching plans. The interactivity and participation are relatively weak, and students are often in a state of passively accepting knowledge, which to a certain extent suppresses their interest and enthusiasm in learning.

[0004] In addition, traditional online education systems cannot provide customized learning paths and recommendations based on students' individual differences and learning progress. This may cause students to encounter some difficult-to-overcome obstacles in the learning process, thus affecting their learning effectiveness and efficiency. Insufficient interaction between teachers and students is also a shortcoming of traditional online education. In traditional online education, the interaction between teachers and students is often limited by time and space, and students may not be able to get timely guidance and help from teachers when needed.

[0005] For example, when students majoring in environmental art design encounter specific difficulties in the design process, they may not be able to get immediate feedback and guidance from teachers, which may affect their design progress and the quality of their works. At the same time, due to the lack of effective communication channels and feedback mechanisms, it is difficult for teachers to fully understand students' learning situations and problems, and thus cannot provide targeted teaching guidance. Summary of the invention

[0006] The technical problem to be solved by the present invention is to provide an online education learning path recommendation system for environmental art design major, which provides personalized learning path recommendations according to students' characteristics and needs, thereby improving teaching effects.

[0007] In order to solve the above technical problems, the technical solution of the present invention is as follows: The online education learning path recommendation system for environmental art design majors includes: User information acquisition module, used to obtain the user's basic information, learning background, interest preferences and learning goals; The course resource database module is used to store various online course resources of the environmental art design major, including course videos, teaching courseware, practical projects and case analysis; The learning behavior analysis module is used to analyze the behavioral characteristics of users in the learning process, including learning style, learning habits and content preferences, and obtain the user's learning behavior pattern; The learning path analysis module is used to construct the initial learning path based on user data, course resources and behavior patterns, and calculate the quality value of each learning path through weighted summation based on the matching degree between the course and user needs, the interaction between the user and the course, the completion of prerequisites, and the adaptability of the course difficulty to the user's ability. The frog swarm algorithm is used to continuously optimize these learning paths, and iterative updates through local search and global information exchange, ultimately determining a high-quality learning sequence that meets user needs; The recommendation engine module is used to generate personalized learning path recommendation solutions based on the final user learning path and send them to the user; A feedback adjustment module is used to receive user feedback on the learning path recommendation plan, and adjust and optimize the learning path recommendation plan based on the feedback information; The learning process evaluation module is used to evaluate the user's learning progress, learning effect and learning status by collecting the user's learning behavior data, including learning time, task completion efficiency and test results, and generate the user's learning evaluation report.

[0008] Furthermore, the user's behavioral characteristics during the learning process are analyzed, including learning style, learning habits and content preferences, to obtain the user's learning behavior pattern to obtain analysis results, including: Through the user information acquisition module and the interaction records during the learning process, various behavioral data of users when learning the online course of environmental art design are obtained, including users' browsing history, click behavior, learning time, frequency of participating in discussions, and homework submission; Extract features that reflect users’ learning styles, learning habits, and content preferences from various behavioral data; Construct an initial classification model including root nodes, internal nodes and leaf nodes, and input the extracted features into the initial classification model. Through the iterative training process, the classification model parameters are continuously adjusted to obtain the classification model of the final user's learning behavior pattern; The classification model is used to classify the user's current behavior data to obtain the user's learning behavior pattern, including learning style characteristics, learning habit summary and content preference analysis.

[0009] Furthermore, the learning style characteristics include the user's preferred learning method and learning rhythm; the learning habit characteristics include daily learning time period, learning frequency and note-taking habits; and the content preference characteristics include the course topics of interest and the preferred teaching resource types.

[0010] Furthermore, an initial classification model including root nodes, internal nodes and leaf nodes is constructed, and the extracted features are input into the initial classification model. Through the iterative training process, the classification model parameters are continuously adjusted to obtain the classification model of the final user learning behavior pattern, including: The features reflecting learning style, learning habits and content preference extracted from user behavior data are used as training set data, and an empty decision tree is initialized. The decision tree contains only one root node. Select one feature from the extracted features as the split feature of the current node, and determine a split condition for the selected feature; According to the splitting condition, the data of the current node is divided into two or more subsets, each subset corresponds to a child node; For each child node, repeat the process of feature selection and split condition setting, and recursively build the subtree structure of the decision tree until the preset stop condition is reached to form the initial decision tree model; The initial decision tree model is used as the initial classification model, which includes the basic structure of the decision tree, namely the root node, internal nodes and leaf nodes; The initial classification model is trained and optimized using the training data. Through an iterative training process, the initial classification model parameters are continuously adjusted, including splitting conditions and tree structure, to obtain a classification model of the final user's learning behavior pattern.

[0011] Furthermore, based on user data, course resources and behavior patterns, the initial learning path is constructed, and the quality value of each learning path is calculated through weighted summation according to the matching degree between the course and user needs, the interaction between the user and the course, the completion of the prerequisites, and the adaptability of the course difficulty to the user's ability. The frog swarm algorithm is used to continuously optimize these learning paths, and iterative updates are made through local search and global information exchange to finally determine a high-quality learning sequence that meets user needs, including: Based on user data, online course resources and user learning behavior patterns, a set of initial learning paths are generated as individuals in the frog population. Each learning path is represented as a sequence of course resources, and the order in the sequence represents the user's learning order. Set the size and initial parameters of the frog group, and calculate the quality value of each learning path through weighted summation based on the match between the course and user needs, the interaction between the user and the course, the completion of prerequisites, and the adaptability of the course difficulty to the user's ability; The frog group is divided into several subgroups, each of which contains a certain number of frogs. In each subgroup, a local search operation is performed to obtain a set of optimized learning paths. After the local search, the global information exchange operation is performed to obtain the result of the global information exchange, and the positions of the frogs in the frog group are updated according to the result of the global information exchange; The operations of local search, global information exchange and frog group update are repeated until the preset number of iterations is reached, and the final solution is determined from the frog group according to the path quality value of the frogs, and the final solution is used as the final user learning path.

[0012] Furthermore, the calculation formula of the path quality value of each learning path is: ; in, Indicates the quality value of the learning path; , , , represents the weight coefficient; represents the course feature vector; Represents the user demand feature vector; Indicates the number of courses the user has been exposed to; Represents the course index; Indicates user With Courses The number of interactions; Indicates user Total number of interactions; Indicates the course The popularity weight of Indicates the number of prerequisites that the user has completed; Indicates the total number of prerequisites required for the course; Indicates the difficulty level of the course; Indicates the user's ability level; Indicates difficulty range.

[0013] Furthermore, based on the final user learning path, a personalized learning path recommendation plan is generated and sent to the user, including: According to the final user learning path, the similarity between users is calculated, and the similarity between courses is analyzed to determine the degree of association between courses, and the analysis results of user similarity and course similarity are obtained; Generate a list of recommended courses based on the analysis results of user similarity and course similarity; Based on the recommended course list, the user's learning goals, schedule and prerequisite factors are analyzed to generate a personalized learning path recommendation plan, including the recommended course sequence, course introduction, learning duration and expected completion time, and the learning path recommendation plan is sent to the user through the learning platform messaging system.

[0014] Furthermore, the calculation formula for the similarity between users is: ; in, Indicates user and users The similarity between Indicates user The eigenvector of Indicates user The eigenvector of , , represents the weight coefficient; Indicates user A collection of items of interest; Indicates user A collection of items of interest; Indicates user and users Overlapping study time; Indicates user Total study time; Indicates user Total study time; Indicates user A collection of projects to participate in learning; Indicates user A collection of projects for participating in learning.

[0015] Furthermore, receiving user feedback on the learning path recommendation plan, and adjusting and optimizing the learning path recommendation plan based on the feedback information, including: Obtain user feedback data, including user account, recommendation account, feedback type and feedback content, Analyze user feedback data and extract key information, including user evaluation of content difficulty, suggestions for learning sequence, and feedback on resource quality, to form feedback analysis results; According to the feedback analysis results, the user feature vector is adjusted, the correlation analysis between courses is optimized, and the content difficulty assessment is adjusted to generate a new learning path recommendation plan for each user.

[0016] Furthermore, by collecting the user's learning behavior data, including learning time, task completion efficiency and test results, the user's learning progress, learning effect and learning status are evaluated, and a learning evaluation report for the user is generated, including: Set the number of training rounds and learning rate for the deep learning model, and collect user learning behavior data, including learning time, task completion efficiency, and test results; Input the learning time and the task completion status into the deep learning model for training to obtain a trained deep learning model; Use the trained deep learning model to classify the user's learning progress, obtain the percentage of completed courses and the estimated remaining time, and analyze the learning time and task completion speed trends to identify learning patterns and active periods, generate learning status monitoring results including learning stability and enthusiasm, and analyze test results to evaluate learning effects, including test scores and knowledge points mastered; Integrate the learning progress assessment results, learning status monitoring results and learning effect evaluation to generate a learning assessment report for the user.

[0017] The above solution of the present invention includes at least the following beneficial effects: Through the user information acquisition module, the system can fully understand the user's basic information, learning background, interest preferences and learning goals, laying a solid foundation for the recommendation of personalized learning paths. The learning behavior analysis module further analyzes the user's learning style, learning habits and content preferences, making the recommended learning path more in line with the user's actual needs, greatly improving the personalization and pertinence of learning. The course resource database module integrates various online course resources of environmental art design majors, including course videos, teaching courseware, practical projects and case analysis, providing users with rich and diverse learning resources. The learning path analysis module intelligently analyzes and determines the final user learning path based on user data and course resources, ensuring that users can use these resources efficiently and orderly, avoiding blind learning and waste of resources.

[0018] The recommendation engine module generates a personalized learning path recommendation plan based on the final user learning path and sends it to the user to guide the user to learn along a scientific and reasonable path. Through this personalized learning path, users can master knowledge points faster, improve learning efficiency, and ensure the improvement of learning results. The feedback adjustment module receives user feedback on the learning path recommendation plan, and adjusts and optimizes the learning path recommendation plan based on the feedback information. This continuous optimization and adjustment mechanism enables the learning path recommendation plan to continuously adapt to user changes and needs, maintaining its timeliness and accuracy.

[0019] The learning process evaluation module conducts a comprehensive evaluation of the user's learning progress, learning effect and learning status by collecting the user's learning behavior data, including learning time, task completion efficiency and test results.

[0020] The generated learning assessment report provides users with clear and intuitive learning feedback, helping them understand their learning status and progress trajectory. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a schematic diagram of a learning path recommendation system for online education of environmental art design major provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0023] like Figure 1 As shown, the embodiment of the present invention proposes an online education learning path recommendation system for environmental art design major, including: The user information acquisition module 11 is used to obtain the user's basic information, learning background, interest preferences and learning goals; The course resource database module 12 is used to store various online course resources of the environmental art design major, including course videos, teaching courseware, practical projects and case analysis; The learning behavior analysis module 13 is used to analyze the behavioral characteristics of the user during the learning process, including learning style, learning habits and content preference, and obtain the user's learning behavior pattern; The learning path analysis module 14 is used to construct an initial learning path based on user data, course resources and behavior patterns, and calculate the quality value of each learning path through weighted summation according to the matching degree between the course and user needs, the interaction between the user and the course, the completion of prerequisites, and the adaptability of the course difficulty to the user's ability, and continuously optimize these learning paths using the frog swarm algorithm, and iterate and update through local search and global information exchange to finally determine a high-quality learning sequence that meets user needs; The recommendation engine module 15 is used to generate a personalized learning path recommendation plan based on the final user learning path and send it to the user; A feedback adjustment module 16 is used to receive user feedback on the learning path recommendation scheme, and adjust and optimize the learning path recommendation scheme according to the feedback information; The learning process evaluation module 17 is used to evaluate the user's learning progress, learning effect and learning status by collecting the user's learning behavior data, including learning time, efficiency of completing tasks and test results, and generate a learning evaluation report for the user.

[0024] In the embodiment of the present invention, by comprehensively acquiring the user's basic information, learning background, interest preferences and learning goals, the system can build a detailed profile of the user and provide accurate basic data for subsequent personalized learning path recommendations. This helps the system better understand the user's needs and expectations, thereby providing learning resources and suggestions that are more in line with the user's actual situation.

[0025] The course resource database module 12 centrally stores various online course resources of the environmental art design major, including course videos, teaching courseware, practical projects and case analysis, providing users with a rich and diverse learning resource library. This enables users to choose the learning resources that suit them at any time according to their needs and interests, thereby improving the flexibility and autonomy of learning.

[0026] The learning behavior analysis module 13 can accurately identify the user's learning style, learning habits and content preferences by conducting in-depth analysis of the user's behavioral characteristics during the learning process. This helps the system provide users with more personalized learning path recommendations and improve the pertinence and efficiency of learning.

[0027] The learning path analysis module 14 combines user data, online course resources and user learning behavior patterns, and the system can intelligently analyze and determine the final user learning path. This ensures the scientificity and rationality of the learning path, helps users to study in an orderly manner according to the established path, and improves the systematicness and coherence of learning.

[0028] The recommendation engine module 15 generates a personalized learning path recommendation plan based on the final user learning path and sends it to the user in real time. This provides users with convenient and instant learning guidance, helps users to grasp their own learning direction and progress at any time, and improves their initiative and enthusiasm in learning.

[0029] The feedback adjustment module 16 receives user feedback on the learning path recommendation scheme, so that the system can timely understand the user's needs and opinions, and adjust and optimize the learning path recommendation scheme according to the feedback information.

[0030] This enables the system to continuously adapt to user changes and needs, maintaining the timeliness and accuracy of learning path recommendations.

[0031] The learning process evaluation module 17 collects the user's learning behavior data, including learning time, task completion efficiency and test results, so that the system can comprehensively evaluate the user's learning progress, learning effect and learning status. The generated learning evaluation report provides users with objective and accurate learning feedback, helps users understand their own learning situation and progress trajectory, and provides a strong reference and guidance for subsequent learning. At the same time, it also provides data support for the system to optimize the learning path recommendation plan.

[0032] In a preferred embodiment of the present invention, the basic information, learning background, interest preferences and learning goals of the user are obtained; various online course resources of the environmental art design major are stored, including course videos, teaching courseware, practical projects and case analysis, which may include: In an embodiment of the present invention, a user information collection form is designed, which should include the user's basic information (such as name, age, gender, etc.), learning background (such as education, major, relevant work experience, etc.), interest preferences (such as favorite design style, design field of interest, etc.) and learning goals (such as skills you want to master, expected learning outcomes, etc.). Develop user registration and login functions to ensure that users can safely and conveniently register an account and log in to the system. During the registration process, the user needs to fill in the above-designed information collection form. The information filled in by the user is stored in the database for subsequent analysis and use. The database should choose a secure and stable storage solution, such as a relational database or a NoSQL database. Preprocess the user information, such as cleaning, deduplication, formatting, etc., to ensure the accuracy and consistency of the data.

[0033] Develop a user information display page so that users can view and edit their own information. At the same time, provide a user information privacy setting function to allow users to control the visibility of their own information. First, classify and organize the online course resources of the environmental art design major, including course videos, teaching courseware, practical projects, and case studies. Establish a unified naming specification and storage format for each type of resource to ensure the standardization and manageability of the resources. Develop a course resource management system that should support operations such as uploading, downloading, editing, and deleting resources. The system should have permission management functions to ensure that only authorized users can perform corresponding operations on resources. Choose a suitable storage solution to store course resources, such as cloud storage, local server storage, etc. Ensure the reliability, security, and scalability of the storage solution. Back up course resources regularly to prevent data loss or damage.

[0034] Develop a course resource search function to allow users to quickly find the required resources by keywords, categories, tags, etc. Design a course resource display page to display resource information in an intuitive and easy-to-use way, such as resource name, description, preview image, download link, etc. Establish a course resource update mechanism to regularly update and add new resources to ensure the timeliness and integrity of resources. Regularly review and maintain existing resources, delete outdated or invalid resources, and optimize resource quality and user experience.

[0035] In a preferred embodiment of the present invention, the user's behavioral characteristics during the learning process are analyzed, including learning style, learning habits and content preferences, and the user's learning behavior pattern is obtained to obtain the analysis results, which may include: Through the user information acquisition module and the interaction records during the learning process, various behavioral data of users when learning the online course of environmental art design are obtained, including users' browsing history, click behavior, learning time, frequency of participating in discussions, and homework submission; Extract features that reflect users’ learning styles, learning habits, and content preferences from various behavioral data; Construct an initial classification model including root nodes, internal nodes and leaf nodes, and input the extracted features into the initial classification model. Through the iterative training process, the classification model parameters are continuously adjusted to obtain the classification model of the final user's learning behavior pattern; The classification model is used to classify the user's current behavior data to obtain the user's learning behavior pattern, including learning style characteristics, learning habit summary and content preference analysis.

[0036] In the embodiment of the present invention, the basic information and background of the user are collected through the user information acquisition module.

[0037] When users study online courses in environmental art design, the system logs are used to record the user's interactive behaviors, including browsing records (such as pages visited and dwell time), click behaviors (such as buttons and links clicked), learning time, frequency of participating in discussions (such as number of posts and replies), and homework submission (such as number of homework submissions and grades). From the collected behavioral data, features that reflect the user's learning style, learning habits, and content preferences are extracted. For example, learning style can be inferred by the user's browsing speed and click frequency (fast browsing may indicate a preference for fast learning, and frequent clicks may indicate more attention to details).

[0038] Study habits can be analyzed by the distribution of study time and the time periods of discussion participation. Content preferences can be identified by the types of pages users visit and the content of the links they click.

[0039] Construct an initial classification model with root nodes, internal nodes, and leaf nodes, such as a decision tree or random forest. Take the extracted features as input, and continuously adjust the parameters of the classification model, such as splitting criteria and pruning strategies, through an iterative training process. Use methods such as cross-validation to evaluate the performance of the model and select the optimal model as the classification model for the end-user's learning behavior pattern. Use the trained classification model to classify the user's current behavior data. Output the user's learning behavior pattern, including learning style characteristics (such as fast learners, detail-oriented), learning habit summary (such as night learners, regular discussion participants), and content preference analysis (such as preference for interior design, landscape design).

[0040] Assume that a user is studying an online course in environmental art design, and the system records the following behavior data: Browsing history: Visited 10 interior design pages, with an average stay of 2 minutes on each page.

[0041] Click behavior: Clicked 20 times on design detail images and 5 times on design theory links.

[0042] Study time: The total study time is 3 hours, 2 hours of which are from 9pm to 11pm.

[0043] Discussion participation: Participated in 3 design discussions, with each post or reply exceeding 500 words.

[0044] Assignment submission: Submitted 2 interior design assignments, both with excellent grades.

[0045] The following features are extracted: Browsing speed: Slow (average 2 minutes / page).

[0046] Click preference: preference for design details (images with more clicks).

[0047] Study time: Night learners (the study time accounts for a high proportion between 9pm and 11pm).

[0048] Discussion activity: active participant (posting or replying many words).

[0049] Homework performance: Excellent (good homework grades).

[0050] These features are input into the trained classification model, and the model outputs the user's learning behavior pattern: Learning Style: Detail-focused (prefers to see design details).

[0051] Study habits: Night learner (study for long hours at night).

[0052] Content preference: Interior design (pages visited and assignments submitted are all about interior design).

[0053] By analyzing the user's learning behavior patterns, the system can provide more personalized learning resource recommendations and learning path planning to meet the user's unique needs. After understanding the user's learning style and habits, the system can adjust the presentation and order of learning content so that users can master knowledge points faster and improve learning efficiency. By analyzing the user's content preferences, the system can recommend design fields and cases that the user is interested in, stimulate the user's learning interest and motivation, and thus optimize the learning effect. The user behavior data and analysis results collected by the system can provide teachers with teaching feedback, help teachers understand students' learning situation and needs, and thus improve teaching methods and content. Based on the user's learning behavior pattern, the system can predict the user's future learning behavior and needs, prepare corresponding learning resources and support in advance, and guide users to continue learning and development.

[0054] In another preferred embodiment of the present invention, the learning style features include the user's preferred learning method and learning rhythm; the learning habit features include daily learning time period, learning frequency and note-taking habits; the content preference features include the course topics of interest and the preferred teaching resource types, which may include: In an embodiment of the present invention, the user prefers a learning method, such as visual learning (preferring charts and videos), auditory learning (preferring lectures and podcasts), or hands-on learning (preferring experiments and projects).

[0055] Learning pace: The user's learning speed, such as fast learning (completing a large amount of content in a short period of time), steady learning (moderate speed, focusing on understanding) or slow learning (meticulous and takes time to digest). Through the user's behavior records on the learning platform, such as the videos, charts, and lecture pages visited, the number of experiments and projects participated in, and the time to complete the course. Use user feedback to collect users' self-reported learning style and pace preferences. Analyze the proportion of videos, charts, lectures and other resources visited by users to determine the user's learning style preference. Calculate the time required for users to complete a course or task, compare it with the average completion time, and determine the user's learning pace. Quantify learning style and pace preferences as numerical values ​​or category labels. Store these features in the user behavior database and associate them with the user ID.

[0056] Define the characteristics of study habits: Daily study time: The time period when users tend to study every day, such as morning, afternoon, or evening.

[0057] Study frequency: the number of times a user studies per day or per week.

[0058] Note-taking habits: whether users take notes, and the level of detail of their notes. Use system logs to record the time when users log in, start and end learning, and the duration of their learning sessions. Analyze whether users create notes during the learning process, as well as the content and quality of their notes. Count the number of learning sessions of users in different time periods every day to determine the daily learning time period. Count the number of learning sessions of users every day or week to determine the learning frequency. Analyze the frequency and content of user notes to determine note-taking habits. Quantify the daily learning time period, learning frequency, and note-taking habits into numerical values ​​or category labels. Store these features in the user behavior database and associate them with the user ID. The courses or knowledge fields preferred by users, such as interior design, landscape design, and architectural design. The teaching resource forms preferred by users, such as video tutorials, PDF documents, and practical projects. Record the courses or resources that users browse, click, collect, and comment on. Determine their areas of interest through the course discussions and homework that users participate in. Count the courses or resource topics that users browse, click, collect, and comment on to determine the course topics of interest. Analyze the types of resources used in the course discussions, homework, or projects that users participate in to determine the preferred teaching resource types. Quantify the course topics of interest and the types of teaching resources preferred as numerical values ​​or category labels. Store these features in the user behavior database and associate them with the user ID.

[0059] The extracted learning style features, learning habit features, and content preference features are integrated into a user behavior feature vector. A user learning behavior pattern classification model is constructed using a decision tree. The model is trained using the user behavior feature vector, and the model parameters are adjusted to optimize the classification performance. Finally, a model that can accurately classify user learning behavior patterns is obtained, which is used for real-time analysis of the user's current behavior data and personalized recommendations.

[0060] In another preferred embodiment of the present invention, an initial classification model including a root node, internal nodes and leaf nodes is constructed, and the extracted features are input into the initial classification model. Through an iterative training process, the classification model parameters are continuously adjusted to obtain a classification model of the final user's learning behavior pattern, which may include: The features reflecting learning style, learning habits and content preference extracted from user behavior data are used as training set data, and an empty decision tree is initialized. The decision tree contains only one root node. Select one feature from the extracted features as the split feature of the current node, and determine a split condition for the selected feature; According to the splitting condition, the data of the current node is divided into two or more subsets, each subset corresponds to a child node; For each child node, repeat the process of feature selection and split condition setting, and recursively build the subtree structure of the decision tree until the preset stop condition is reached to form the initial decision tree model; The initial decision tree model is used as the initial classification model, which includes the basic structure of the decision tree, namely the root node, internal nodes and leaf nodes; The initial classification model is trained and optimized using the training data. Through an iterative training process, the initial classification model parameters are continuously adjusted, including splitting conditions and tree structure, to obtain a classification model of the final user's learning behavior pattern.

[0061] In an embodiment of the present invention, features such as learning style (such as learning method, learning rhythm), learning habits (such as daily learning time period, learning frequency, note-taking habits) and content preferences (such as interested course topics, preferred teaching resource types) are extracted from the user behavior database. These features are combined into feature vectors as training set data. In the process of building a decision tree, a feature is selected from the feature set of the current node as a split feature. For the selected split feature, a split condition is determined, such as a certain threshold or category of the feature. This condition will be used to divide the data of the current node into different subsets. The split condition is applied to the data of the current node to divide the data into two or more subsets that meet the condition and do not meet the condition. A new child node is created for each subset, and the subset data is assigned to the corresponding child node. For each newly created child node, the process of feature selection, split condition setting and data division is repeated. This process is recursively performed until a preset stop condition is reached, such as the maximum depth of the tree, the minimum number of samples of the node, or the threshold of the split gain. When the stop condition is met, the recursion is stopped to form an initial decision tree model. The constructed initial decision tree model is saved as an initial classification model. The model contains the basic structure of a decision tree, including the root node, internal nodes (including split features and conditions), and leaf nodes (representing classification results or categories). The initial classification model is trained using the training set data. Through the iterative training process, the model parameters are continuously adjusted, such as split conditions, tree structure (such as pruning, merging nodes), etc. The performance of the model is evaluated using methods such as cross-validation and accuracy evaluation, and the model is optimized based on the evaluation results. When the model performance reaches a satisfactory level, the final user learning behavior pattern classification model is obtained.

[0062] Suppose there is a user learning behavior dataset that contains the following features: learning method (visual, auditory, hands-on practice), learning rhythm (fast, steady, slow), daily learning time period (morning, afternoon, evening), course topics of interest (interior design, landscape design, architectural design), etc.

[0063] Create an empty decision tree with only one root node. Select "Learning Style" as the splitting feature of the root node. If the learning style is "Vision", it is divided into the left child node; otherwise, it is divided into the right child node. According to the splitting condition, the data is divided into two subsets: "Vision" learning style and "Non-Vision" learning style. Create a new child node for each subset and assign the subset data to the corresponding child node. For the left child node, select "Learning Pace" as the splitting feature, and determine the splitting condition as follows: if the learning pace is "Fast", it is divided into the left child node; otherwise, it is divided into the right child node. For the right child node (non-visual learning style), select "Course Topic of Interest" as the splitting feature, and determine the splitting condition as follows: if the course topic of interest is "Interior Design", it is divided into the left child node; otherwise, it is divided into the right child node. This process is performed recursively until the preset stopping condition is reached, such as the maximum depth of the tree is 3. Save the constructed initial decision tree model, which contains the root node, internal nodes, and leaf nodes. Train the initial classification model using the training set data. Through the iterative training process, the model parameters are adjusted, such as pruning to simplify the tree structure, merging similar nodes, etc. The model performance is evaluated and the model is optimized based on the evaluation results. Finally, a classification model of user learning behavior patterns is obtained.

[0064] By analyzing the user's learning behavior pattern, the classification model can provide users with personalized learning resource recommendations and learning path planning to meet the user's unique needs and improve learning efficiency and effectiveness. The decision tree model is interpretable and can clearly display the classification process and decision path, helping to understand how the model is classified according to user behavior characteristics. Through the iterative training process and continuous adjustment of model parameters, the performance of the model can be optimized, and the classification accuracy and generalization ability can be improved. The classification model can provide teachers with teaching feedback to help teachers understand students' learning behavior and preferences, thereby improving teaching methods and content and improving teaching quality. Based on the classification model, users' future learning behaviors and needs can be predicted, providing decision support for teaching institutions, while helping users better plan and manage their own learning process.

[0065] In a preferred embodiment of the present invention, an initial learning path is constructed based on user data, course resources and behavior patterns, and the quality value of each learning path is calculated by weighted summation according to the matching degree between the course and user needs, the interaction between the user and the course, the completion of the prerequisites, and the adaptability of the course difficulty to the user's ability. The frog swarm algorithm is used to continuously optimize these learning paths, and iterative updates are made through local search and global information exchange to finally determine a high-quality learning sequence that meets user needs, which may include: Based on user data, online course resources and user learning behavior patterns, a set of initial learning paths are generated as individuals in the frog population. Each learning path is represented as a sequence of course resources, and the order in the sequence represents the user's learning order. Set the size and initial parameters of the frog group, and calculate the quality value of each learning path through weighted summation based on the match between the course and user needs, the interaction between the user and the course, the completion of prerequisites, and the adaptability of the course difficulty to the user's ability; The frog group is divided into several subgroups, each of which contains a certain number of frogs. In each subgroup, a local search operation is performed to obtain a set of optimized learning paths. After the local search, the global information exchange operation is performed to obtain the result of the global information exchange, and the positions of the frogs in the frog group are updated according to the result of the global information exchange; The operations of local search, global information exchange and frog group update are repeated until the preset number of iterations is reached, and the final solution is determined from the frog group according to the path quality value of the frogs, and the final solution is used as the final user learning path.

[0066] In an embodiment of the present invention, the user's basic information, learning style, learning habits, content preferences and other features are extracted from the user database. A list of online course resources is obtained, including attributes such as course name, difficulty level, and pre-requirements. A set of initial learning paths is randomly generated in combination with user characteristics and course resource attributes. Each learning path is a sequence composed of course resources, and the course sequence in the sequence represents the learning sequence recommended by the user. These initial learning paths are used as individuals in a frog group, and each individual represents a possible learning path solution. The size of the frog group is set, that is, the number of frogs (learning paths) in the frog group. Initialize the parameters of the frog group, such as the range of local search, the frequency of global information exchange, etc. Calculate the path quality value for each learning path, which can be evaluated based on factors such as the matching degree of course resources, the coherence of the learning path, and the degree of satisfaction of user preferences. The frog group is randomly or according to a certain rule divided into several subgroups, each of which contains a certain number of frogs (learning paths). In each subgroup, a local search operation is performed on the frog (learning path). Local search can be performed by exchanging course resources in the learning path, adjusting the course sequence, etc., to find a better learning path. After local search, a set of optimized learning paths are obtained, which have higher path quality values ​​within the subgroup.

[0067] After the local search is completed, the global information exchange operation is performed. Global information exchange can be carried out through frog exchange between subgroups, broadcasting of the global optimal solution, etc., to promote information sharing and cooperation in the frog group. According to the results of the global information exchange, the position of the frogs in the frog group is updated. That is, according to the global final solution or the final solution within the subgroup, the course resource composition and order of the frogs (learning paths) in the frog group are adjusted. The number of iterations is set, that is, the number of times the local search, global information exchange and frog group update operations are repeated. In each iteration, the local search, global information exchange and frog group update operations are performed in sequence to continuously optimize the learning path in the frog group. When the preset number of iterations is reached, the final solution is determined from the frog group according to the frog's path quality value. The final solution is the learning path with the highest path quality value. The final solution is recommended to the user as the final user learning path.

[0068] Suppose there is a user whose learning style is visual, whose learning habit is to study every night, and whose content preference is courses related to interior design. Online course resources include basic courses, advanced courses and project practice courses, each of which has difficulty levels and prerequisites. Generate a set of initial learning paths based on user characteristics and course resource attributes. For example, path 1: [basic course A, advanced course B, project practice course C]; path 2: [basic course D, advanced course E, project practice course F], etc. Set the frog group size to 10, that is, it contains 10 learning paths. Initialize parameters, such as the local search range is 2 (that is, each local search can exchange or adjust 2 course resources), and the global information exchange frequency is once every 5 iterations. Divide the frog group into 2 subgroups, each of which contains 5 learning paths. Perform local search in each subgroup, and optimize the learning path by exchanging or adjusting course resources. For example, path 1 in subgroup 1 becomes [basic course A, project practice course C, advanced course B] after local search. In the fifth iteration, a global information exchange operation is performed to broadcast the optimal path in subgroup 1 (such as [basic course A, project practice course C, advanced course B]) to subgroup 2. Based on the results of the global information exchange, the learning paths in the frog group are updated. For example, a path in subgroup 2 may refer to the optimal path in subgroup 1 to adjust its own course resource composition and order. After multiple iterations, the final solution is determined from the frog group based on the path quality value. Assume that the final solution is: [basic course A, advanced course G, project practice course C], which has the highest path quality value. The final solution is recommended to the user as the final user learning path.

[0069] By combining user data, online course resources, and user learning behavior patterns, we can provide users with personalized learning path recommendations to meet their unique needs and learning preferences. Through local search and global information exchange operations in the frog swarm algorithm, we can continuously optimize the quality of learning paths and improve the consistency and effectiveness of learning paths. Personalized learning paths can guide users to learn in a reasonable order and with reasonable resources, avoiding blind trial and error and wasting time, thereby improving learning efficiency. By providing users with customized learning path recommendations, we can enhance users' learning experience and satisfaction, and improve users' stickiness and loyalty to the learning platform. Analyzing users' learning paths and preferences can provide decision support for teaching institutions, help them optimize course design and teaching resource allocation, and improve teaching quality and effectiveness.

[0070] In a preferred embodiment of the present invention, the calculation formula of the path quality value of each learning path is: ; in, Indicates the quality value of the learning path; , , , represents the weight coefficient; represents the course feature vector; Represents the user demand feature vector; Indicates the number of courses the user has been exposed to; Represents the course index; Indicates user With Courses The number of interactions; Indicates user Total number of interactions; Indicates the course The popularity weight of Indicates the number of prerequisites that the user has completed; Indicates the total number of prerequisites required for the course; Indicates the difficulty level of the course; Indicates the user's ability level; Indicates difficulty range.

[0071] In the embodiment of the present invention, the course feature vector is defined as and user demand feature vector These two vectors can contain multiple dimensions, such as course type, content area, teaching style, user interests, learning goals, etc. Initialize weight coefficients , , , These coefficients are used to adjust the path quality values ​​of each part of the formula Initialize the number of courses the user has been exposed to , Course Index , the number of interactions between users and courses , the total number of user interactions , Course popularity weight , the number of prerequisites that the user has completed , the total number of prerequisites required by the course , Course Difficulty Level , user capability level , and difficulty range Using the dot product formula Calculate the similarity between the course feature vector and the user demand feature vector. Calculate the modulus of the course feature vector. and the model of the user demand feature vector According to the formula Calculate the standardized matching degree between courses and user needs. Traverse all courses that the user has been exposed to, and for each course , count the number of interactions between the user and the course Total number of user interactions The ratio, multiplied by the course Popularity weight , get the contribution of the course to the quality of user interaction. Average the contribution of all courses to get the overall quality of user-course interaction. Count the number of prerequisites that the user has completed Calculates the number of prerequisites a user has completed relative to the total number of prerequisites required for the course The proportion .

[0072] Calculate course difficulty level User capability level The absolute difference between . Use difficulty range Normalize this difference to get the mismatch between the course difficulty and the user's ability. Subtract the mismatch from 1 to get the match between the course difficulty and the user's ability. Calculate the quality value of the learning path according to the formula . Continuously optimize the quality value of the learning path , and obtain the final user learning path.

[0073] By comprehensively considering course characteristics, user needs, user interaction behaviors, prerequisite completion, and the matching degree between course difficulty and user ability, more personalized and accurate learning path recommendations can be provided to users. The optimized learning path can better meet the user's learning needs and interests, reduce the blind trial and error and waste of time in the learning process, and thus improve learning efficiency. Personalized learning paths can improve the user's learning experience and satisfaction, making users more willing to use the learning platform and continue to learn. By analyzing the user's learning path and quality value, teaching institutions can understand the user's learning preferences and needs, provide decision support for course design and teaching resource allocation, and further optimize the teaching quality and effect. The optimized learning path can stimulate the user's interest and motivation in learning, prompt the user to continue learning, form a virtuous circle, and continuously improve their learning ability and level.

[0074] In a preferred embodiment of the present invention, a personalized learning path recommendation scheme is generated according to the final user learning path and sent to the user, which may include: According to the final user learning path, the similarity between users is calculated, and the similarity between courses is analyzed to determine the degree of association between courses, and the analysis results of user similarity and course similarity are obtained; Generate a list of recommended courses based on the analysis results of user similarity and course similarity; Based on the recommended course list, the user's learning goals, schedule and prerequisite factors are analyzed to generate a personalized learning path recommendation plan, including the recommended course sequence, course introduction, learning duration and expected completion time, and the learning path recommendation plan is sent to the user through the learning platform messaging system.

[0075] In an embodiment of the present invention, the learning path data of all users is extracted, including the courses they have been exposed to, the number of interactions, the prerequisites completed, etc. A feature vector is constructed for each user, and the vector dimensions may include the number of courses the user has been exposed to, the number of interactions for a specific course, the proportion of prerequisites completed, etc. The similarity between user feature vectors is calculated using cosine similarity. A user similarity matrix is ​​obtained, which represents the degree of similarity between each pair of users. The similarity between courses is analyzed to determine the degree of association between courses. The characteristics of the courses, such as course type, content area, difficulty level, popularity, etc., are extracted.

[0076] The similarity between the course feature vectors is calculated using a similarity calculation method (such as cosine similarity) to obtain a course similarity matrix, which represents the similarity between each pair of courses. ;in, , Represents the course feature vector. According to the course similarity matrix, the degree of correlation between courses is determined, such as high correlation, medium correlation or low correlation.

[0077] For the target user, find several users with the highest similarity to the target user, analyze the learning paths of these users, and extract the courses that they have been exposed to but the target user has not been exposed to. For the courses that the target user has been exposed to, find several courses with the highest similarity to the target user as recommendation candidates. Combine the analysis results of user similarity and course similarity to generate a preliminary list of recommended courses. Sort and filter the recommended courses according to factors such as the popularity, difficulty level, and prerequisites of the courses to obtain the final list of recommended courses. Analyze the learning goals of the target user to determine whether the recommended courses meet their learning goals. According to the learning habits and available time of the target user, reasonably arrange the order and learning duration of the recommended courses. Check the prerequisites of the recommended courses to ensure that the target user has met or can learn these prerequisites at the same time. Based on the above analysis, generate a personalized learning path recommendation plan, including the recommended course order, course introduction, learning duration, and estimated completion time. Ensure that the messaging system of the learning platform works properly and can send personalized messages. Send the generated learning path recommendation plan to the target user through the messaging system to remind the user to check and start learning.

[0078] Assume that the learning path of target user A mainly includes data structure and algorithm basic courses, and he shows a strong interest in the field of machine learning during the learning process. Find users B and C who are most similar to user A, and find that in addition to learning data structure and algorithm basic courses, they also learned machine learning and deep learning courses. Analyze the data structure and algorithm basic courses that user A has been exposed to, and find the courses with the highest similarity, such as Introduction to Machine Learning and Basic Deep Learning. Combine the analysis results of user similarity and course similarity to generate a list of recommended courses: Introduction to Machine Learning, Basic Deep Learning, and Introduction to Reinforcement Learning. Analyze user A's learning goals (master basic machine learning skills), time arrangement (can devote 10 hours of study per week), and prerequisites (have mastered the basics of data structure and algorithm).

[0079] Generate a recommended plan: first learn the introduction to machine learning (estimated study time 4 weeks), then learn the basics of deep learning (estimated study time 4 weeks), and finally learn the introduction to reinforcement learning (estimated study time 2 weeks). Send the recommended plan to user A through the learning platform message system, reminding him to start learning in the recommended order.

[0080] By combining user similarity and course similarity, learning paths that meet user interests and needs can be recommended more accurately. Personalized learning paths can reduce users' blind trial and error in the learning process, improve learning efficiency, and help users achieve their learning goals faster. Accurate learning path recommendations can enhance users' learning experience and satisfaction, increase users' stickiness to the learning platform, and promote users' continuous learning. By analyzing users' learning paths and recommending courses, teaching institutions can understand users' learning needs and preferences, and provide decision support for course design and teaching resource allocation. Personalized learning path recommendations can encourage users to explore and learn new areas, promote the dissemination and sharing of knowledge, and promote academic and technological progress.

[0081] In a preferred embodiment of the present invention, the calculation formula for the similarity between users is: ; in, Indicates user and users The similarity between Indicates user The eigenvector of Indicates user The eigenvector of , , represents the weight coefficient; Indicates user A collection of items of interest; Indicates user A collection of items of interest; Indicates user and users Overlapping study time; Indicates user Total study time; Indicates user Total study time; Indicates user A collection of projects to participate in learning; Indicates user A collection of projects for participating in learning.

[0082] In the embodiment of the present invention, the user is calculated and users The eigenvector of and This involves multiplying the corresponding components of the two vectors and then adding the results. The dot product reflects the degree of similarity between the two feature vectors in direction and is the basis of similarity calculation. and users Collection of Interest Items and The intersection and union of the two users are shown in Figure 1. The size of the intersection represents the number of items that the two users are interested in, and the size of the union represents the total number of items that the two users are interested in. The ratio of the intersection size divided by the union size is used to measure the similarity between the two users in terms of their interest items.

[0083] Identify users and users Overlapping study time This can be found by comparing the learning time periods of two users. and users Total study time and The similarity between two users in their learning time is measured by dividing the overlap in learning time by the shorter total learning time of the two users. and users A collection of projects for participating in learning and The size of the intersection reflects the number of items that two users participate in learning together, and is an important indicator to measure the similarity of their learning behaviors.

[0084] The results of the above four parts are calculated according to the given weight coefficients. , , The dot product of the feature vector directly reflects the feature similarity. The similarity of the interest project set is measured by the intersection ratio. The overlap of learning time is measured by the ratio of the overlap time to the shorter total learning time. The similarity of the participating learning project set is measured by the size of the intersection. Finally, the result of the weighted sum is divided by the product of the modulus of the feature vector (i.e. ), to standardize the similarity value. As a user and users The similarity measure between .

[0085] User characteristics, interest projects, learning time, and participation in learning projects can more comprehensively measure the similarity between users. , , , the focus of similarity calculation can be customized according to specific application scenarios and needs. Comprehensive consideration of multiple factors can improve the accuracy of similarity calculation, allowing the recommendation system to more accurately identify similar users, thereby improving the recommendation effect. By considering personalized factors such as user interest projects, learning time, and participation in learning projects, more personalized recommendations and services can be provided to users.

[0086] In a preferred embodiment of the present invention, receiving user feedback on the learning path recommendation scheme, and adjusting and optimizing the learning path recommendation scheme according to the feedback information may include: Obtain user feedback data, including user account, recommendation account, feedback type and feedback content, Analyze user feedback data and extract key information, including user evaluation of content difficulty, suggestions for learning sequence, and feedback on resource quality, to form feedback analysis results; According to the feedback analysis results, the user feature vector is adjusted, the correlation analysis between courses is optimized, and the content difficulty assessment is adjusted to generate a new learning path recommendation plan for each user.

[0087] In an embodiment of the present invention, a feedback collection system is set up, which can receive and store feedback data submitted by users. When a user submits feedback, the system automatically records the user account, the recommended solution account (i.e., the solution identifier when the user receives the recommendation), the feedback type (such as satisfaction, dissatisfaction, suggestions, etc.) and the feedback content (the user's specific opinions or suggestions). Ensure that the feedback data is stored in a structured manner to facilitate subsequent analysis and processing. Use natural language processing technology (NLP) to pre-process the feedback content, including removing irrelevant words and punctuation marks, and performing word segmentation and part-of-speech tagging. Apply text mining techniques, such as keyword extraction and sentiment analysis, to extract key information from the feedback content. Specifically, identify the user's evaluation of the difficulty of the content (such as "too difficult", "moderate", "too simple"), suggestions for the learning order (such as "should learn X before learning Y"), and feedback on the quality of resources (such as "the video is clear" and "the explanation is not clear"). The extracted key information is organized into feedback analysis results to provide a basis for subsequent adjustments and optimizations. Analyze the user preferences and needs mentioned in the feedback, such as the user's preference for certain types of courses and expectations for the learning rhythm. Based on this information, adjust the corresponding dimensions in the user feature vector to more accurately reflect the user's current status and needs. For example, if user feedback indicates that they prefer practical courses, then the weights associated with practical courses in the user feature vector can be increased. Re-evaluate the order and dependencies between courses based on the user's suggestions for the learning sequence. Update the course relevance matrix to reflect the new information in the user feedback. For example, if the user recommends taking course A before taking course B, then the relevance of course A to course B can be increased. Adjust the evaluation criteria for course difficulty levels based on the user's evaluation of content difficulty. If most users think a course is too difficult, consider lowering its difficulty level or providing it with supplementary learning materials. On the contrary, if a course is considered too simple, consider increasing its difficulty or providing more advanced course options. Rerun the learning path recommendation algorithm using the updated user feature vector, course relevance matrix, and content difficulty evaluation criteria. Generate a new learning path recommendation for each user that better meets their needs and preferences.

[0088] Suppose user A submits a feedback with the following content: "I think course X is too difficult. I should take course Y before X. Moreover, the video explanation is a bit unclear." Receive and store feedback data: user account A, recommended solution account (assuming it is solution 1), feedback type (suggestion), feedback content ("Course X is too difficult. I should take course Y before X. The video explanation is not clear").

[0089] Analyze the feedback content: Extract key information, such as "Course X is too difficult", "Learn Y before X", "The video explanation is not clear".

[0090] Adjust the user feature vector: increase user A’s preference weight for practical courses (assuming Y is a practical course) and reduce his acceptance of courses that are too difficult.

[0091] Optimize course relevance analysis: Increase the relevance of course Y to course X, which means that learning Y before X is more in line with the user's learning sequence.

[0092] Adjust content difficulty assessment: reduce the difficulty level of course X, or provide supplementary learning materials for it; at the same time, mark courses where the video explanation is unclear for subsequent improvement.

[0093] Generate a new learning path recommendation: Based on the updated information, generate a new learning path for user A, where course Y is ranked before course X, and clear video explanations or auxiliary materials are provided.

[0094] By obtaining and analyzing user feedback, we can more accurately understand user needs and preferences, thereby generating learning path recommendations that better meet user expectations. Personalized learning path recommendation solutions can meet the different needs of users and improve their learning experience and satisfaction. By continuously optimizing and adjusting learning path recommendation solutions, users can be encouraged to continue learning and improve learning efficiency and results. The ability to adjust and optimize recommendation solutions in real time based on user feedback shows good adaptability and flexibility, and can better respond to changes in user needs and behaviors. User feedback on course content and quality can provide valuable improvement suggestions for course developers, which helps to continuously improve the quality of courses and teaching effectiveness.

[0095] In a preferred embodiment of the present invention, by collecting the user's learning behavior data, including learning time, task completion efficiency and test results, the user's learning progress, learning effect and learning status are evaluated, and a learning evaluation report for the user is generated, which may include: Set the number of training rounds and learning rate for the deep learning model, and collect user learning behavior data, including learning time, task completion efficiency, and test results; Input the learning time and the task completion status into the deep learning model for training to obtain a trained deep learning model; Use the trained deep learning model to classify the user's learning progress, obtain the percentage of completed courses and the estimated remaining time, and analyze the learning time and task completion speed trends to identify learning patterns and active periods, generate learning status monitoring results including learning stability and enthusiasm, and analyze test results to evaluate learning effects, including test scores and knowledge points mastered; Integrate the learning progress assessment results, learning status monitoring results and learning effect evaluation to generate a learning assessment report for the user.

[0096] In an embodiment of the present invention, the architecture of the deep learning model is determined, such as using a convolutional neural network (CNN), and the most appropriate model is selected according to the characteristics of the learning behavior data. The number of training rounds (epochs) of the model, that is, the number of times the model iterates on the training data set, and the learning rate (learning rate) are set to control the step size of the model to update the weights in each iteration. A data collection system is deployed to collect users' learning behavior data in real time or regularly, including learning time, efficiency of completing tasks (such as task completion time, accuracy, etc.) and test results (such as test scores, wrong answers, etc.). The collected learning time and task completion data are preprocessed, such as data cleaning, normalization or standardization, to ensure the quality and consistency of the data.

[0097] Input the preprocessed data into the deep learning model to start the training process. The model gradually adjusts its weights and parameters by learning patterns and associations in the data. During the training process, monitor the model's loss function and accuracy to evaluate the model's training effect. Use the trained deep learning model to classify the user's learning progress and predict the percentage of the user's completed course and the estimated remaining time. Analyze the user's learning time and task completion speed trends to identify the user's learning patterns and active periods. For example, through time series analysis, find the user's peak and trough in learning. Generate learning status monitoring results, including learning stability and enthusiasm. Stability can be evaluated by the fluctuation of the user's learning time, and enthusiasm can be reflected by the speed and efficiency of task completion. Perform a detailed analysis of the user's test results, including test scores, wrong answers, and knowledge points mastered. Use statistical analysis to identify which knowledge points the user performs well and which knowledge points need further improvement. Generate a learning effect evaluation report that clearly points out the user's learning strengths and weaknesses, as well as the direction for improvement. Integrate the learning progress assessment results, learning status monitoring results and learning effect assessment to generate a learning assessment report for the user. The report should include the user's learning progress, learning status, learning effect, and targeted suggestions and improvement measures. Present the learning assessment report to the user in a visual or textual form to facilitate user understanding and self-adjustment.

[0098] Assume that user Xiao Ming has been studying on an online learning platform for a period of time, and data on his learning behavior is collected and evaluated. Xiao Ming's learning time, efficiency of completing tasks (such as the time and accuracy of each task), and test results (such as unit test scores) are collected. A deep learning model is trained using the collected data, and an appropriate number of training rounds and learning rate are set. The trained model classifies Xiao Ming's learning progress and predicts that he has completed 80% of the course and the remaining time is expected to be 2 weeks. By analyzing the trend of learning time and task completion speed, it is found that Xiao Ming is more active in learning on weekends and evenings, and the learning time is relatively stable. The learning status monitoring results are generated, indicating that Xiao Ming's learning stability is good and his enthusiasm is high. After analyzing Xiao Ming's test results, it is found that he has a good grasp of the knowledge points in data structure, but needs to be further strengthened in algorithm design. The above evaluation results are integrated into a learning evaluation report, which clearly points out Xiao Ming's learning progress, learning status, learning effect, and the direction for improvement.

[0099] By generating learning assessment reports for users, we can provide users with personalized learning support and suggestions, helping them to better understand their learning situation and room for improvement. Learning assessment reports can point out users' learning weaknesses and areas that need improvement, helping users to adjust their learning strategies in a targeted manner and improve their learning efficiency. By showing users' learning progress and results, learning assessment reports can enhance users' learning motivation and self-confidence, and motivate users to continue learning. Based on users' learning assessment reports, we can optimize the allocation of teaching resources, such as providing more tutoring materials and practice questions for users who need extra help. The data and analysis results in learning assessment reports can provide valuable feedback for educators, helping them understand teaching effectiveness and students' learning needs, so as to continuously improve teaching methods and content.

[0100] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. The online education learning path recommendation system for environmental art design major is characterized by: include: User information acquisition module, used to obtain the user's basic information, learning background, interest preferences and learning goals; The course resource database module is used to store various online course resources for the environmental art design major, including course videos, teaching courseware, practical projects and case analysis; The learning behavior analysis module is used to analyze the behavioral characteristics of users in the learning process, including learning style, learning habits and content preferences, and obtain the user's learning behavior pattern; The learning path analysis module is used to construct the initial learning path based on user data, course resources and behavior patterns, and calculate the quality value of each learning path through weighted summation based on the matching degree between the course and user needs, the interaction between the user and the course, the completion of prerequisites, and the adaptability of the course difficulty to the user's ability. The frog swarm algorithm is used to continuously optimize these learning paths, and iterative updates through local search and global information exchange, ultimately determining a high-quality learning sequence that meets user needs; The recommendation engine module is used to generate personalized learning path recommendation solutions based on the final user learning path and send them to the user; A feedback adjustment module is used to receive user feedback on the learning path recommendation plan, and adjust and optimize the learning path recommendation plan based on the feedback information; The learning process evaluation module is used to evaluate the user's learning progress, learning effect and learning status by collecting the user's learning behavior data, including learning time, task completion efficiency and test results, and generate the user's learning evaluation report.

2. The online education learning path recommendation system for environmental art design major according to claim 1 is characterized in that: Analyze the user's behavioral characteristics during the learning process, including learning style, learning habits and content preferences, obtain the user's learning behavior pattern, and obtain analysis results, including: Through the user information acquisition module and the interaction records during the learning process, various behavioral data of users when learning the online course of environmental art design are obtained, including users' browsing history, click behavior, learning time, frequency of participating in discussions, and homework submission; Extract features that reflect users’ learning styles, learning habits, and content preferences from various behavioral data; Construct an initial classification model including root nodes, internal nodes and leaf nodes, and input the extracted features into the initial classification model. Through the iterative training process, the classification model parameters are continuously adjusted to obtain the classification model of the final user's learning behavior pattern; The classification model is used to classify the user's current behavior data to obtain the user's learning behavior pattern, including learning style characteristics, learning habit summary and content preference analysis.

3. The online education learning path recommendation system for environmental art design major according to claim 2 is characterized in that: Learning style characteristics include the user's preferred learning method and learning rhythm; learning habit characteristics include daily learning time period, learning frequency and note-taking habits; content preference characteristics include interested course topics and preferred teaching resource types.

4. The online education learning path recommendation system for environmental art design major according to claim 3 is characterized in that: Construct an initial classification model containing root nodes, internal nodes, and leaf nodes, and input the extracted features into the initial classification model. Through the iterative training process, continuously adjust the classification model parameters to obtain the classification model of the end user's learning behavior pattern, including: The features reflecting learning style, learning habits and content preference extracted from user behavior data are used as training set data, and an empty decision tree is initialized. The decision tree contains only one root node. Select one feature from the extracted features as the split feature of the current node, and determine a split condition for the selected feature; According to the splitting condition, the data of the current node is divided into two or more subsets, each subset corresponds to a child node; For each child node, repeat the process of feature selection and split condition setting, and recursively build the subtree structure of the decision tree until the preset stop condition is reached to form the initial decision tree model; The initial decision tree model is used as the initial classification model, which includes the basic structure of the decision tree, namely the root node, internal nodes and leaf nodes; The initial classification model is trained and optimized using the training data. Through an iterative training process, the initial classification model parameters are continuously adjusted, including splitting conditions and tree structure, to obtain a classification model of the final user's learning behavior pattern.

5. The online education learning path recommendation system for environmental art design major according to claim 4 is characterized in that: Based on user data, course resources and behavior patterns, the initial learning path is constructed. The quality value of each learning path is calculated through weighted summation based on the matching degree between the course and user needs, the interaction between the user and the course, the completion of the prerequisites, and the adaptability of the course difficulty to the user's ability. The frog swarm algorithm is used to continuously optimize these learning paths. Through local search and global information exchange, iterative updates are made to finally determine a high-quality learning sequence that meets user needs, including: Based on user data, online course resources and user learning behavior patterns, a set of initial learning paths are generated as individuals in the frog population. Each learning path is represented as a sequence of course resources, and the order in the sequence represents the user's learning order. Set the size and initial parameters of the frog group, and calculate the quality value of each learning path through weighted summation based on the match between the course and user needs, the interaction between the user and the course, the completion of prerequisites, and the adaptability of the course difficulty to the user's ability; The frog group is divided into several subgroups, each of which contains a certain number of frogs. In each subgroup, a local search operation is performed to obtain a set of optimized learning paths. After the local search, the global information exchange operation is performed to obtain the result of the global information exchange, and the positions of the frogs in the frog group are updated according to the result of the global information exchange; The operations of local search, global information exchange and frog group update are repeated until the preset number of iterations is reached, and the final solution is determined from the frog group according to the path quality value of the frogs, and the final solution is used as the final user learning path.

6. The online education learning path recommendation system for environmental art design major according to claim 5 is characterized in that: The calculation formula for the path quality value of each learning path is: ； in, Indicates the quality value of the learning path; , , , represents the weight coefficient; represents the course feature vector; Represents the user demand feature vector; Indicates the number of courses the user has been exposed to; Represents the course index; Indicates user With Courses The number of interactions; Indicates user Total number of interactions; Indicates the course The popularity weight of Indicates the number of prerequisites that the user has completed; Indicates the total number of prerequisites required for the course; Indicates the difficulty level of the course; Indicates the user's ability level; Indicates difficulty range.

7. The online education learning path recommendation system for environmental art design major according to claim 6 is characterized in that: Generate a personalized learning path recommendation plan based on the final user learning path and send it to the user, including: According to the final user learning path, the similarity between users is calculated, and the similarity between courses is analyzed to determine the degree of association between courses, and the analysis results of user similarity and course similarity are obtained; Generate a list of recommended courses based on the analysis results of user similarity and course similarity; Based on the recommended course list, the user's learning goals, schedule and prerequisite factors are analyzed to generate a personalized learning path recommendation plan, including the recommended course sequence, course introduction, learning duration and expected completion time, and the learning path recommendation plan is sent to the user through the learning platform messaging system.

8. The online education learning path recommendation system for environmental art design major according to claim 7 is characterized in that: The formula for calculating the similarity between users is: ； in, Indicates user and users The similarity between Indicates user The eigenvector of Indicates user The eigenvector of , , represents the weight coefficient; Indicates user A collection of items of interest; Indicates user A collection of items of interest; Indicates user and users Overlapping study time; Indicates user Total study time; Indicates user Total study time; Indicates user A collection of projects to participate in learning; Indicates user A collection of projects for participating in learning.

9. The online education learning path recommendation system for environmental art design major according to claim 8 is characterized in that: Receive user feedback on the learning path recommendation plan, and adjust and optimize the learning path recommendation plan based on the feedback information, including: Obtain user feedback data, including user account, recommendation account, feedback type and feedback content, analyze user feedback data, and extract key information, including user evaluation of content difficulty, suggestions for learning sequence, and feedback on resource quality, to form feedback analysis results; According to the feedback analysis results, the user feature vector is adjusted, the correlation analysis between courses is optimized, and the content difficulty assessment is adjusted to generate a new learning path recommendation plan for each user.

10. The online education learning path recommendation system for environmental art design major according to claim 9 is characterized in that: By collecting the user's learning behavior data, including learning time, task completion efficiency and test results, the user's learning progress, learning effect and learning status are evaluated, and the user's learning evaluation report is generated, including: Set the number of training rounds and learning rate for the deep learning model, and collect user learning behavior data, including learning time, task completion efficiency, and test results; Input the learning time and the task completion status into the deep learning model for training to obtain a trained deep learning model; Use the trained deep learning model to classify the user's learning progress, obtain the percentage of completed courses and the estimated remaining time, and analyze the learning time and task completion speed trends to identify learning patterns and active periods, generate learning status monitoring results including learning stability and enthusiasm, and analyze test results to evaluate learning effects, including test scores and knowledge points mastered; Integrate the learning progress assessment results, learning status monitoring results and learning effect evaluation to generate a learning assessment report for the user.

Images