Recommendation methods and related equipment for artificial intelligence teaching resources in primary and secondary schools

Abstract

The present invention relates to a method for recommending artificial intelligence teaching resources for primary and secondary schools and related equipment, comprising the following steps: analyzing the learning behavior of target users through educational data mining technology to obtain learning behavior characteristics; using a preset TransE algorithm, based on the learning behavior characteristics, performing association mapping on knowledge points in preset educational resources to obtain mapped teaching knowledge points; using a preset natural language processing technology, analyzing the difficulty of the mapped teaching knowledge points, and screening the mapped teaching knowledge points based on the difficulty to obtain screened mapped teaching knowledge points; performing structured modeling on the screened mapped teaching knowledge points to obtain a knowledge system map; and recommending teaching resources based on the knowledge system map through an intelligent decision-making algorithm, thereby solving the technical problem that traditional educational resource recommendation systems often rely on teachers' experience and personal judgment and are difficult to meet large-scale personalized teaching needs.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method for recommending artificial intelligence teaching resources for primary and secondary schools and related equipment. Background Art

[0002] With the rapid development of information technology, the application of artificial intelligence in the field of education has gradually deepened, providing new possibilities for personalized teaching. Especially for primary and secondary education, the rational use of artificial intelligence technology can effectively improve the quality of teaching and students' learning efficiency. However, in practical applications, due to the significant differences in students' cognitive levels and interest preferences, how to accurately provide teaching resources that meet individual needs has become an urgent problem to be solved. Traditional educational resource recommendation systems often rely on teachers' experience and personal judgment, and it is difficult to meet the needs of large-scale personalized teaching.

[0003] In addition, the existing teaching resource recommendation system also has certain limitations in the recommendation algorithm. For example, some recommendation systems mainly rely on the user's browsing history or simple tag matching to recommend resources. This recommendation method lacks analysis of students' deep learning behavior, resulting in insufficient accuracy and pertinence of the recommendation results. At the same time, the artificial intelligence teaching resource recommendation system for primary and secondary school students also faces the problem of mismatch between the difficulty of knowledge points and students' existing cognitive level, which not only affects the learning effect, but may also dampen students' enthusiasm for learning. Therefore, it is particularly important to develop a teaching resource recommendation system that can comprehensively consider students' learning behavior characteristics and knowledge difficulty.

[0004] In order to overcome the above challenges, this study proposed an artificial intelligence teaching resource recommendation method based on educational data mining technology and natural language processing. This method first uses educational data mining technology to deeply analyze the learning behavior of target users to obtain more refined and comprehensive learning behavior characteristics; then uses the TransE algorithm to associate and map the knowledge points in the teaching resources to construct a more scientific knowledge system map; finally, combined with the intelligent decision-making algorithm, the most appropriate teaching resources are recommended based on the students' learning characteristics and knowledge mastery. This method can not only improve the accuracy of the recommendation system, but also promote the development of students' personalized learning, and contribute to the realization of high-quality educational equity. Summary of the invention

[0005] The main purpose of the present invention is to provide a method and related equipment for recommending artificial intelligence teaching resources for primary and secondary schools, which solves the technical problem that traditional educational resource recommendation systems often rely on teachers' experience and personal judgment and are difficult to meet the needs of large-scale personalized teaching.

[0006] To achieve the above object, the present invention provides a method for recommending artificial intelligence teaching resources for primary and secondary schools, comprising the following steps: Analyze the learning behavior of target users through educational data mining technology to obtain learning behavior characteristics; Using the preset TransE algorithm, based on the learning behavior characteristics, the knowledge points in the preset educational resources are associated and mapped to obtain mapped teaching knowledge points; Analyzing the difficulty of the mapped teaching knowledge points through a preset natural language processing technology, and screening the mapped teaching knowledge points based on the difficulty to obtain screened mapped teaching knowledge points; Performing structured modeling on the filtered mapped teaching knowledge points to obtain a knowledge system map; Through an intelligent decision-making algorithm, teaching resources are recommended based on the knowledge system map.

[0007] Furthermore, the learning behavior analysis of the target users is performed by using educational data mining technology to obtain learning behavior characteristics, including: By using a preset statistical analysis method, historical learning data of the target user is analyzed to obtain a learning activity curve; wherein the learning activity curve includes the learning activity and learning time distribution of the target user, and the learning activity curve is a curve with the learning activity as the Y-axis and the learning time distribution as the X-axis; Classifying the learning behavior pattern of the target user based on the learning activity curve to obtain a learning feature pattern; Based on the learning feature pattern, the target user's emotion change is identified to obtain an emotion change trend; wherein the emotion change trend is the target user's emotion state at different learning stages, and the emotion change trend includes a positive state, a neutral state, and a negative state; Through educational data mining technology, historical learning data mining is performed on the target user based on the emotion change trend to obtain learning behavior characteristics; wherein the learning behavior characteristics include interest preferences and learning habits.

[0008] Furthermore, the preset TransE algorithm is used to perform association mapping on the knowledge points in the preset educational resources based on the learning behavior characteristics to obtain mapped teaching knowledge points, including: Using a semantic network construction algorithm, the learning behavior characteristics are converted into semantic network structure data; Based on a word vector algorithm, vector mapping is performed on the learning behavior features in the semantic network to obtain a mapping feature vector; Determining the weight of the mapping feature vector according to the importance of the learning behavior feature through a vector weighting algorithm to obtain a learning behavior weight vector; Converting the knowledge points in the educational resource into vector format to obtain knowledge point vectors; By using a preset cosine similarity algorithm, the angle between the knowledge point vector and the learning behavior weight vector in the vector space is calculated to obtain a cosine similarity value between the knowledge point vector and the learning behavior weight vector; Sorting the knowledge points corresponding to the knowledge point vectors according to the cosine similarity values ​​to obtain a sorted knowledge point sequence; The preset TransE algorithm is used to map the sorted knowledge point sequence into a low-dimensional vector space to obtain mapped teaching knowledge points.

[0009] Furthermore, the difficulty of the mapped teaching knowledge points is analyzed by a preset natural language processing technology, and the mapped teaching knowledge points are screened based on the difficulty to obtain the screened mapped teaching knowledge points, including: By using a preset semantic ambiguity model, it is determined whether the mapping teaching knowledge point has semantic ambiguity. If so, the mapping teaching knowledge point is semantically clarified by using a preset semantic disambiguation technology to obtain semantically clear knowledge points; Constructing a dependency syntax tree for the semantically clear knowledge point to obtain a knowledge point with a dependency syntax tree; By using the semantic role labeling module in the semantic ambiguity model, the knowledge points with the dependency syntax tree are labeled with semantic roles to obtain the knowledge points labeled with semantic roles; Using a preset semantic network analysis algorithm, the semantic importance of the knowledge points labeled with the semantic roles is evaluated to obtain a knowledge point importance vector; Analyzing the degree of difficulty of the knowledge point importance vector by the cluster analysis module in the semantic network analysis algorithm, and clustering the knowledge point importance vector based on the degree of difficulty to obtain a cluster analysis result; wherein the cluster analysis result includes a vector with difficulty clustering; The mapped teaching knowledge points are screened based on the cluster analysis results through a preset natural language processing technology to obtain screened mapped teaching knowledge points.

[0010] Furthermore, the structured modeling of the filtered mapped teaching knowledge points to obtain a knowledge system map includes: Performing semantic coding on the filtered mapped teaching knowledge points to obtain semantically coded knowledge points; Constructing a semantic network for the semantically encoded knowledge points to obtain a semantic relationship knowledge point network; The ontology framework is constructed for the semantic relationship knowledge point network by using a preset ontology construction technology to obtain a knowledge point network with an ontology framework; wherein the knowledge point network with an ontology framework includes nodes and edges, wherein the nodes are knowledge points and the edges are relationships between knowledge points; The matching degree between the nodes and the edges is measured by a preset energy function. If the matching degree is not within a preset matching range, the hierarchical structure and relationship constraints of the knowledge points are obtained, and the hierarchical updating of the knowledge point network is performed based on the hierarchical structure and relationship constraints by a preset hierarchical clustering algorithm to obtain a knowledge point network after hierarchical updating; semantically grouping the knowledge point network after the hierarchical update to obtain a semantically grouped knowledge point network; Performing relationship combing on the semantically grouped knowledge point network to obtain a relationship combing knowledge network; Through graph fusion technology, the relationships are organized into a knowledge network for data fusion to obtain a knowledge system graph.

[0011] Furthermore, the intelligent decision-making algorithm is used to recommend teaching resources based on the knowledge system map, including: Based on the knowledge system graph, the learning behavior characteristics are graph matched to obtain a user learning behavior characteristic graph; The similarity between the learning behavior feature graph and the knowledge system graph is calculated by a preset graph similarity calculation algorithm to obtain a similarity matrix; Using a preset graph clustering algorithm, clustering the knowledge system graph based on the similarity matrix to obtain multiple knowledge point clusters; Based on the learning behavior characteristic graph and each of the knowledge point clusters, planning a learning path for the target user to obtain a learning path graph; Using a preset resource recommendation algorithm, based on the learning path map and the learning behavior characteristics, recommend preliminary teaching resources that match the current learning status of the target user; The feedback information of the target users on the recommended preliminary teaching resources is collected through a preset feedback mechanism, and the preliminary teaching resources are dynamically adjusted based on the feedback information to obtain teaching resources.

[0012] Furthermore, based on the learning behavior characteristic graph and the knowledge point clustering, a learning path of a target user is planned to obtain a learning path graph, including: Step S71, determining the learning path starting point of the target user based on the learning behavior characteristic graph by using a preset path starting point selection algorithm; Step S72, predicting the learning path endpoint of the target user based on the path starting point and the knowledge point clustering by using a preset path endpoint prediction algorithm; Step S73, generating a preliminary learning path based on the path starting point and the path end point; Step S74, based on the preliminary learning path and the knowledge point clustering, optimizing the order of path nodes in the preliminary learning path to obtain an optimized learning path; Step S75, verifying the optimized learning path to determine whether the optimized learning path is reasonable and coherent, and obtaining a verification result; Step S76, if the verification result is qualified, then the optimized learning path is visualized to obtain a learning path diagram; In step S77, if the verification result is not met, the learning behavior characteristic graph is adjusted according to the verification result, and the process returns to step S71 to re-determine the starting point of the learning path of the target user based on the adjusted learning behavior characteristic graph until the optimized learning path is reasonable and coherent.

[0013] The present invention also provides a recommendation system for primary and secondary school artificial intelligence teaching resources, including: The analysis module is used to analyze the learning behavior of target users through educational data mining technology to obtain learning behavior characteristics; A mapping module, used to use a preset graph embedding algorithm to perform association mapping on the knowledge points in the preset educational resources based on the learning behavior characteristics to obtain mapped teaching knowledge points; A screening module, used to analyze the difficulty of the mapping teaching knowledge points through a preset natural language processing technology, and screen the mapping teaching knowledge points based on the difficulty to obtain screened mapping teaching knowledge points; A modeling module, used to perform structured modeling on the filtered mapped teaching knowledge points to obtain a knowledge system map; The recommendation module is used to recommend teaching resources based on the knowledge system map through an intelligent decision-making algorithm.

[0014] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the above methods when executing the computer program.

[0015] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above-mentioned methods are implemented.

[0016] The method for recommending artificial intelligence teaching resources for primary and secondary schools provided by the present invention comprises the following steps: analyzing the learning behavior of target users by educational data mining technology to obtain learning behavior characteristics; using a preset TransE algorithm, based on the learning behavior characteristics, performing association mapping on knowledge points in preset educational resources to obtain mapped teaching knowledge points; analyzing the difficulty of the mapped teaching knowledge points by a preset natural language processing technology, and screening the mapped teaching knowledge points based on the difficulty to obtain screened mapped teaching knowledge points; performing structured modeling on the screened mapped teaching knowledge points to obtain a knowledge system map; recommending teaching resources based on the knowledge system map by an intelligent decision-making algorithm; through the above-mentioned technical means, the problem that traditional educational resource recommendation systems often rely on teachers' experience and personal judgment and are difficult to meet the needs of large-scale personalized teaching is solved; based on the constructed knowledge system map and intelligent decision-making algorithm, the system can dynamically adjust the recommended content according to the learning progress and mastery of each student, truly realizing the intelligent recommendation of personalized teaching resources and promoting the implementation of the concept of teaching students in accordance with their aptitude. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a schematic diagram of the steps of a method for recommending artificial intelligence teaching resources for primary and secondary schools in one embodiment of the present invention; Figure 2 It is a structural block diagram of a recommendation system for artificial intelligence teaching resources for primary and secondary schools in one embodiment of the present invention; Figure 3 It is a schematic block diagram of the structure of a computer device according to an embodiment of the present invention.

[0018] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings in conjunction with the embodiments. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0020] like Figure 1 As shown, Figure 1 It is a schematic diagram of steps of a method for recommending artificial intelligence teaching resources for primary and secondary schools in one embodiment of the present invention; In one embodiment of the present invention, a method for recommending artificial intelligence teaching resources for primary and secondary schools is provided, comprising the following steps: Step S1, analyzing the learning behavior of target users through educational data mining technology to obtain learning behavior characteristics.

[0021] Specifically, the process of obtaining learning behavior characteristics by analyzing the learning behavior of target users is the basic link of the entire recommendation method, which involves the application of educational data mining technology. Specifically, this process first requires the collection of a large amount of learning data from users, which may include but is not limited to multi-dimensional information such as students' clickstream records on the learning platform, homework completion, test scores, online learning time, and interactive feedback. Once the data is collected, the next step is to use educational data mining technology to process and analyze these raw data, with the aim of extracting key features that can reflect students' learning behavior patterns. For example, by analyzing the time and accuracy of students' homework submission, we can understand students' learning habits and their mastery of knowledge points; for example, by observing students' help-seeking behavior when encountering difficult problems, we can evaluate students' problem-solving ability and attitude towards autonomous learning. Furthermore, these learning behavior characteristics will serve as an important basis for the subsequent steps of using the TransE algorithm to map knowledge points, ensuring that the recommended educational resources not only match the students' current learning status, but also effectively promote the development of students' abilities. For example, in a primary and secondary school artificial intelligence teaching resource recommendation system, if data analysis shows that a student performs well in math problem solving but lacks in programming logic thinking, the system will give priority to recommending teaching materials that can both consolidate math advantages and make up for programming shortcomings, thereby achieving more personalized teaching resource recommendations. Such a design can not only improve students' learning efficiency, but also help stimulate their interest in learning and promote all-round development. In addition, the target users include but are not limited to primary and secondary school students.

[0022] Step S2, using the preset TransE algorithm, based on the learning behavior characteristics, the knowledge points in the preset educational resources are associated and mapped to obtain mapped teaching knowledge points.

[0023] Specifically, after completing the analysis of the learning behavior characteristics of the target users, the next step is to use the preset TransE algorithm to associate and map the knowledge points in the preset educational resources based on these learning behavior characteristics, so as to obtain the mapped teaching knowledge points. The TransE algorithm is a knowledge graph embedding method based on the vector space model. It can effectively convert the entities and relationships in the knowledge graph into low-dimensional vectors, and then calculate the similarity or distance between entities, which is of great significance for discovering and constructing potential connections between knowledge points. In this process, the learning behavior characteristics are encoded into vector form together with the knowledge points in the educational resources as part of the input, so that the correlation between them can be identified by calculating the similarity between the vectors of different knowledge points. For example, on a programming learning platform for primary and secondary school students, it is assumed that through the previous learning behavior analysis, it has been known that a student has a good foundation in graphical programming, but has a weak understanding of conditional statements. At this time, using the TransE algorithm, relevant knowledge points such as "graphical programming" and "conditional statements" can be mapped into the same vector space, and their correlation strength can be determined by calculating the vector distance between these knowledge points. If the vector of "conditional statement" is closer to the vector of "loop structure" and farther from the vector of "variable assignment", it can be inferred that there is a strong logical connection between "conditional statement" and "loop structure", while the connection with "variable assignment" is relatively weak. Based on such mapping results, the recommendation system can more intelligently recommend teaching resources containing exercises related to "conditional statement" and "loop structure" to this student, rather than repeatedly providing materials on "variable assignment", so as to more effectively supplement students' knowledge gaps and promote the overall improvement of their programming skills.

[0024] Step S3, analyzing the difficulty of the mapped teaching knowledge points through a preset natural language processing technology, and screening the mapped teaching knowledge points based on the difficulty to obtain screened mapped teaching knowledge points.

[0025] Specifically, after using the TransE algorithm to complete the associative mapping of knowledge points in educational resources, the next step is to analyze the difficulty of these mapped teaching knowledge points through the preset natural language processing technology, and screen the knowledge points based on the difficulty, and finally obtain the screened mapped teaching knowledge points. This process mainly involves two key operations: one is to use natural language processing technology to evaluate the difficulty of knowledge points, and the other is to effectively screen according to the evaluation results. Natural language processing technology can infer the difficulty of knowledge points by analyzing the vocabulary complexity, sentence structure complexity and the relationship between knowledge points in the description of knowledge points. For example, knowledge points that contain more professional terms and long and difficult sentences are generally considered to be more difficult, while knowledge points with concise and clear language expression are relatively easy to understand. Taking programming learning as an example, suppose that after mapping the two knowledge points of "conditional statements" and "loop structures", the system finds through natural language processing technology that the description of "conditional statements" contains more basic concepts, such as "if-else" structure and "logical operator", and the sentence structure is relatively simple, so it is judged to be of low difficulty; while the description of "loop structure" involves multiple concepts such as "for loop", "while loop" and "loop control statement", and the language expression is more complex, so it is judged to be of high difficulty. Based on such difficulty analysis, the system can further select teaching knowledge points suitable for the current cognitive level of specific students. For example, for beginners, the system will give priority to recommending learning materials related to "conditional statements", and leave "loop structure" as advanced learning content. This can not only avoid students from feeling frustrated because of facing too difficult knowledge points, but also ensure that they can master new knowledge step by step, so as to achieve the best learning effect.

[0026] Step S4, performing structured modeling on the screened mapped teaching knowledge points to obtain a knowledge system map.

[0027] Specifically, after completing the difficulty analysis and screening of the mapping teaching knowledge points, the next step is to perform structured modeling on these screened mapping teaching knowledge points to obtain a knowledge system map. This process aims to organize the scattered knowledge points according to their inherent logical relationships and hierarchical structures to form a clear and intuitive knowledge framework. By constructing a knowledge system map, it can not only help students better understand and remember knowledge points, but also provide teachers with an effective teaching guidance tool to facilitate them to design course content and teaching activities according to the knowledge structure in the map. Specifically, in the process of structured modeling, it is first necessary to determine the logical relationship between knowledge points, such as which knowledge points are basic knowledge, which are advanced knowledge, and which knowledge points have causal relationships or are prerequisites for each other. Then, use data structures such as graph theory or tree structures to represent these relationships to ensure that each knowledge point can find a suitable position in the knowledge field to which it belongs. For example, in the scenario of programming learning, basic knowledge points such as "variables", "conditional statements" and "loop structures" can be placed at the bottom of the knowledge system map, because they are the cornerstones of learning programming languages; and advanced concepts such as "function calls" and "object-oriented programming" can be placed at the upper level of the map, because the mastery of these concepts needs to be based on a full understanding of the basic knowledge. In addition, the map can also clearly mark the direct connection between the knowledge points. For example, "conditional statements" can lead to the part about "loop control" in "loop structure", because these two control structures are often used in combination to solve complex programming problems. Through such structured modeling, the knowledge system map not only provides students with an orderly learning path to help them maintain a sense of direction during the learning process and avoid getting lost in a large number of knowledge points, but also provides teachers with a flexible teaching planning tool, so that they can adjust teaching strategies according to the specific circumstances of students and achieve more personalized teaching goals.

[0028] Step S5, recommending teaching resources based on the knowledge system map through an intelligent decision-making algorithm.

[0029] Specifically, after constructing the knowledge system map, the next step is to recommend teaching resources based on this map through an intelligent decision-making algorithm. The core of this process is to use an intelligent decision-making algorithm to analyze the match between students' personal learning data and the knowledge system map, so as to recommend the most suitable teaching resources for students. The intelligent decision-making algorithm can make comprehensive considerations based on multiple factors, such as students' historical learning records, current learning progress, learning preferences, and their position in the knowledge system map.

[0030] Specifically, when a student logs in to the programming learning platform, the system will first locate his position in the knowledge system map based on the student's previous learning behavior and grades. Assume that the system knows through preliminary analysis that the student has mastered the basic concepts of "variables" and "conditional statements", but needs to be strengthened in "loop structure". At this time, the intelligent decision-making algorithm will refer to the position of the "loop structure" knowledge point in the knowledge system map and its relationship with other knowledge points. For example, "loop structure" is based on "variables" and "conditional statements", and is also a prerequisite for learning advanced concepts such as "arrays" and "function calls". Based on this information, the algorithm will select teaching resources that can help students consolidate existing knowledge and smoothly transition to new knowledge points for recommendation. For example, the system may recommend a series of video tutorials, practical exercises, and interactive case studies on "for loops" and "while loops", all of which are designed to enable students to deepen their understanding of "loop structures" in practice.

[0031] In addition, the intelligent decision-making algorithm will continue to monitor students' feedback and learning outcomes, such as evaluating the effectiveness of recommended resources through indicators such as scores on exercises and participation in online discussions. If it is found that the student is making good progress on the recommended learning path, the system will continue to recommend higher-level learning materials along this path; conversely, if the student encounters difficulties with a certain knowledge point, the algorithm will adjust the recommendation strategy in a timely manner, provide more auxiliary resources or go back to review the basics to ensure that the student can make steady progress. In this way, the intelligent decision-making algorithm can not only improve the personalization of teaching resource recommendations, but also dynamically adapt to students' learning changes, providing each student with the most suitable learning experience.

[0032] In a specific embodiment, the learning behavior analysis of the target user is performed by using educational data mining technology to obtain learning behavior characteristics, including: By using a preset statistical analysis method, historical learning data of the target user is analyzed to obtain a learning activity curve; wherein the learning activity curve includes the learning activity and learning time distribution of the target user, and the learning activity curve is a curve with the learning activity as the Y-axis and the learning time distribution as the X-axis; Classifying the learning behavior pattern of the target user based on the learning activity curve to obtain a learning feature pattern; Based on the learning feature pattern, the target user's emotion change is identified to obtain an emotion change trend; wherein the emotion change trend is the target user's emotion state at different learning stages, and the emotion change trend includes a positive state, a neutral state, and a negative state; Through educational data mining technology, historical learning data mining is performed on the target user based on the emotion change trend to obtain learning behavior characteristics; wherein the learning behavior characteristics include interest preferences and learning habits.

[0033] Specifically, the process of analyzing the learning behavior of target users through educational data mining technology and obtaining learning behavior characteristics is a multi-level and comprehensive analysis process. First, this process involves a detailed statistical analysis of the historical learning data of the target users to extract the time distribution law that can reflect the user's learning activity, that is, to construct a learning activity curve. The learning activity curve here refers to a curve with learning activity as the Y-axis and learning time distribution as the X-axis, which can intuitively show the user's learning activity in different time periods. For example, on a programming learning platform, the system can draw a time curve representing each student's learning activity by recording the time each time the student logs into the learning platform, the online learning time, the number of videos watched, the number of exercises completed, and other data. These curves can help educators or system developers understand the general learning patterns of students, such as whether most students tend to study at night or on weekends, or whether there are certain specific time periods when students' learning activity decreases significantly. Based on these learning activity curves, the system can further classify the learning behavior patterns of target users to obtain learning feature patterns. This means that students with similar learning activity curves are classified into the same group to facilitate the discovery and summary of different learning behavior patterns. For example, the system may identify that one type of students prefers to concentrate on studying in the morning, while another type of students prefers to study in the evening. This classification helps provide more personalized learning advice and support for students with different learning habits. This classification process can be automatically completed through clustering algorithms or other machine learning methods to ensure that each type of learning mode has a certain internal consistency and is clearly different from other categories. After obtaining the learning feature pattern, the system further identifies the emotional changes of the target users to obtain the emotional change trend. The emotional change trend here refers to the emotional state of the target users at different learning stages, including positive, neutral and negative states. This identification process usually requires combining the user's interactive data on the learning platform, such as comments, ratings, and help requests, to determine the user's current emotional tendency through the sentiment analysis algorithm. For example, if a student posts a post with positive comments after completing a programming task, the system can mark it as a positive state; on the contrary, if the student expresses frustration after trying to solve a problem many times without success, it will be marked as a negative state. This identification of emotional change trends plays an important role in timely adjusting teaching strategies, enhancing teacher-student interaction, and improving students' learning motivation and satisfaction. Finally, through educational data mining technology, the system mines historical learning data of target users based on the above emotional change trends, and finally obtains learning behavior characteristics. These learning behavior characteristics include information on interest preferences and learning habits.Interest preference reflects the degree of interest of students in a subject or a specific knowledge point, which can be determined by analyzing data such as the content that students choose to learn, the topics they participate in discussions, and the keywords of search queries. For example, a student who often watches programming video tutorials and actively participates in programming project discussions may indicate that he has a strong interest in programming. Learning habits refer to students' regular practices and tendencies in the learning process, such as a fixed learning time period every day, whether they prefer independent learning or group cooperation, etc. This information can be obtained by analyzing data such as students' login time, learning frequency, and types of activities they participate in. For example, a student who is accustomed to starting to study at 7 o'clock every morning may have good self-discipline and the habit of getting up early to study. In summary, the process of analyzing the learning behavior of target users through educational data mining technology can not only gain an in-depth understanding of students' learning patterns and emotional states, but also tailor a more personalized and effective learning path for each student. For example, on a programming learning platform for primary and secondary school students, the system can arrange more morning courses for students who have high learning efficiency in the morning based on the above analysis results, and for those students who have better learning effects in the evening, more learning resources can be pushed in the evening. At the same time, the system can also provide timely psychological support or learning suggestions based on the students' emotional trends, helping them overcome learning obstacles, maintain a positive learning attitude, and achieve better learning outcomes.

[0034] In a specific embodiment, the preset TransE algorithm is used to associate and map the knowledge points in the preset educational resources based on the learning behavior characteristics to obtain the mapped teaching knowledge points, including: Using a semantic network construction algorithm, the learning behavior characteristics are converted into semantic network structure data; Based on a word vector algorithm, vector mapping is performed on the learning behavior features in the semantic network to obtain a mapping feature vector; Determining the weight of the mapping feature vector according to the importance of the learning behavior feature through a vector weighting algorithm to obtain a learning behavior weight vector; Converting the knowledge points in the educational resource into vector format to obtain knowledge point vectors; By using a preset cosine similarity algorithm, the angle between the knowledge point vector and the learning behavior weight vector in the vector space is calculated to obtain a cosine similarity value between the knowledge point vector and the learning behavior weight vector; Sorting the knowledge points corresponding to the knowledge point vectors according to the cosine similarity values ​​to obtain a sorted knowledge point sequence; The preset TransE algorithm is used to map the sorted knowledge point sequence into a low-dimensional vector space to obtain mapped teaching knowledge points.

[0035] Specifically, the process of using the preset TransE algorithm to associate and map the knowledge points in the preset educational resources based on the learning behavior characteristics to obtain the mapped teaching knowledge points is a highly technical and intelligent operation process. First, this process requires the use of a semantic network construction algorithm to convert the learning behavior characteristics into semantic network structure data. The semantic network here is a data structure that can represent concepts and their relationships. It can capture the semantic connection between learning behavior characteristics, thereby laying the foundation for subsequent vector mapping. For example, in the scenario of programming learning, if the system has identified a student's excellent learning performance in "conditional statements" through educational data mining technology, but is not proficient in "loop structures", then these learning behavior characteristics will be converted into nodes in the semantic network, and the edges between the nodes represent the associations between these features, such as the logical relationship between "conditional statements" and "loop structures". Next, based on the word vector algorithm, the learning behavior characteristics in the semantic network are vector mapped to obtain a mapping feature vector. The word vector algorithm is a technology that converts words into numerical vectors that can maintain the semantic similarity between words in a high-dimensional space. Through this method, learning behavior features such as "conditional statements" and "loop structures" are mapped into corresponding vectors, which not only contain information about the features themselves, but also imply the relationship between the features. For example, the distance between the vectors of "conditional statements" and "loop structures" in space can reflect the relevance of these two knowledge points in the student's learning process. Subsequently, the weight of the mapped feature vector is determined according to the importance of the learning behavior features through the vector weighting algorithm to obtain the learning behavior weight vector. This process is to highlight those features that have a greater impact on students' learning outcomes and ensure that they are given higher weights in subsequent calculations. For example, if the system analyzes that the degree of mastery of "conditional statements" has a greater impact on students' overall programming ability than "loop structures", then when calculating the vector corresponding to "conditional statements", it will be given a greater weight. The purpose of this is to make the recommendation system pay more attention to those knowledge points that have a greater impact on students' learning outcomes. Next, the knowledge points in the educational resources are converted into vector format to obtain knowledge point vectors. This means that the system needs to convert all the knowledge points in the preset educational resources into vector form to facilitate comparison with the learning behavior weight vector. These knowledge point vectors will be used as candidate sets for subsequent similarity calculations. For example, every programming knowledge point in the system, whether it is a basic "variable" definition or a complex "object-oriented programming" concept, will be converted into a vector, and these vectors together constitute a huge knowledge point vector library. Then, through the preset cosine similarity algorithm, the angle between the knowledge point vector and the learning behavior weight vector in the vector space is calculated to obtain the cosine similarity value between the knowledge point vector and the learning behavior weight vector.Cosine similarity is a method to measure the similarity between two vectors. It evaluates their similarity by calculating the angle between the two vectors. The smaller the angle, the more similar the two vectors are. In this example, the system calculates the cosine similarity value between each knowledge point vector and the learning behavior weight vector. The larger the value, the more the knowledge point matches the student's current learning behavior characteristics. The knowledge points corresponding to the knowledge point vectors are sorted according to the cosine similarity value to obtain a sorted knowledge point sequence. This means that the system sorts each knowledge point according to the degree of match between the knowledge point and the learning behavior characteristics to find the knowledge point that best meets the student's current learning needs. For example, after sorting, the system may find that the "loop structure" knowledge point best matches the student's learning behavior characteristics, so it is ranked first in the recommendation list. Finally, the preset TransE algorithm is used to map the sorted knowledge point sequence into a low-dimensional vector space to obtain the mapped teaching knowledge points. The TransE algorithm is a method for knowledge graph embedding that can map vectors in a high-dimensional space into a low-dimensional space while keeping the relative distance between vectors unchanged. Through this method, the system can convert the sorted sequence of knowledge points into a more compact and efficient vector representation, thereby simplifying the recommendation process and improving the efficiency of recommendation. For example, the system may map knowledge points such as "conditional statements" and "loop structures" onto a two-dimensional plane, which not only can intuitively display the relationship between these knowledge points, but also facilitate the subsequent teaching resource recommendation operation. In summary, this series of steps can not only accurately capture the learning behavior characteristics of students, but also efficiently match these characteristics with the knowledge points in educational resources, and finally realize the intelligent recommendation of personalized teaching resources. For example, on the programming learning platform, the system can recommend a series of teaching resources related to "loop structures" to a student who performs well in "conditional statements" but has insufficient mastery of "loop structures", including video tutorials, programming exercises, and case analysis, to help him overcome learning difficulties and achieve comprehensive knowledge mastery.

[0036] In a specific embodiment, the difficulty of the mapped teaching knowledge points is analyzed by a preset natural language processing technology, and the mapped teaching knowledge points are screened based on the difficulty to obtain the screened mapped teaching knowledge points, including: By using a preset semantic ambiguity model, it is determined whether the mapping teaching knowledge point has semantic ambiguity. If so, the mapping teaching knowledge point is semantically clarified by using a preset semantic disambiguation technology to obtain semantically clear knowledge points; Constructing a dependency syntax tree for the semantically clear knowledge point to obtain a knowledge point with a dependency syntax tree; By using the semantic role labeling module in the semantic ambiguity model, the knowledge points with the dependency syntax tree are labeled with semantic roles to obtain the knowledge points labeled with semantic roles; Using a preset semantic network analysis algorithm, the semantic importance of the knowledge points labeled with the semantic roles is evaluated to obtain a knowledge point importance vector; Analyzing the degree of difficulty of the knowledge point importance vector by the cluster analysis module in the semantic network analysis algorithm, and clustering the knowledge point importance vector based on the degree of difficulty to obtain a cluster analysis result; wherein the cluster analysis result includes a vector with difficulty clustering; The mapped teaching knowledge points are screened based on the cluster analysis results through a preset natural language processing technology to obtain screened mapped teaching knowledge points.

[0037] Specifically, the process of analyzing the difficulty of the mapped teaching knowledge points through the preset natural language processing technology and screening the mapped teaching knowledge points based on the difficulty to obtain the mapped teaching knowledge points after screening is a complex operation involving multiple technical links. This process aims to ensure that the teaching resources recommended to students not only match their learning behavior characteristics, but also meet the students' current cognitive level in terms of difficulty, thereby improving the learning effect. First of all, this process needs to use the preset semantic ambiguity model to determine whether there is semantic ambiguity in the mapped teaching knowledge points. Semantic ambiguity means that the same word or sentence may have different meanings in different contexts, and this ambiguity may affect the system's accurate assessment of the difficulty of the knowledge point. For example, in programming learning, the word "table" can refer to both a table in a database and a table in an HTML page, and these two meanings are very different in different contexts. In order to eliminate this ambiguity, the system will use the preset semantic disambiguation technology to semantically clarify the mapped teaching knowledge points, thereby obtaining semantically clear knowledge points. Semantic disambiguation technology is usually based on contextual information and determines the exact meaning of a word by analyzing the vocabulary and sentence structure around the word. For example, if the system detects that the word "table" appears in the context related to "database query", it will interpret it as a table in the database. Next, the dependency syntax tree is constructed for the semantically clear knowledge point to obtain the knowledge point with the dependency syntax tree. The dependency syntax tree is a graph structure used to represent the internal structure of a sentence, which can clearly show the dependency relationship between words. By constructing the dependency syntax tree, the system can understand the internal structure of the knowledge point more deeply and provide support for subsequent semantic analysis. For example, for the knowledge point "use SQL to query the table in the database", the system will construct a dependency syntax tree, in which "use" is the subject-predicate relationship, "SQL" is the tool, "query" is the verb, and "table in the database" is the object. This structure helps the system understand the specific content and logical relationship of the knowledge point. Then, through the semantic role labeling module in the semantic ambiguity model, the knowledge point with the dependency syntax tree is labeled with semantic roles to obtain the knowledge point after semantic role labeling. Semantic role labeling refers to assigning a semantic role to each word in the sentence, such as agent, patient, tool, etc., which helps the system understand the meaning of the sentence more accurately. For example, in the sentence "Use SQL to query the table in the database", "SQL" is marked as a tool role and "table in the database" is marked as an object role. These marking information helps the system identify the key components in the knowledge point. Next, the semantic importance of the knowledge points after semantic role marking is evaluated through the preset semantic network analysis algorithm to obtain the knowledge point importance vector. The semantic network analysis algorithm is a method for analyzing the semantic relationship between words. By constructing a semantic network, the system can evaluate the importance of each knowledge point in the knowledge system.For example, the system may find that the knowledge point "SQL query" is in a central position in the network because it is closely related to multiple other knowledge points (such as "database management" and "data retrieval"), so it will be given a high importance score. Then, through the cluster analysis module in the semantic network analysis algorithm, the difficulty of the knowledge point importance vector is analyzed, and the knowledge point importance vector is clustered based on the difficulty to obtain the cluster analysis results. Cluster analysis is an unsupervised learning method that can group data points with similar characteristics together. In this process, the system will divide the knowledge points into different difficulty levels, such as "easy", "medium" and "difficult" according to the scores of the knowledge point importance vector. This division helps the system recommend teaching resources suitable for students of different levels. For example, for beginners, the system will give priority to recommending knowledge points at the "easy" level, such as "variable definition" and "basic data types", while for advanced learners, it will recommend knowledge points at the "medium" or "difficult" level, such as "function call" and "object-oriented programming". Finally, the mapped teaching knowledge points are screened based on the cluster analysis results through the preset natural language processing technology to obtain the mapped teaching knowledge points after screening. This screening process aims to ensure that the knowledge points recommended to students not only match their learning behavior characteristics, but also meet the students' current cognitive level in terms of difficulty. For example, for a student who performs well in programming learning, the system may recommend more "medium" or "difficult" level knowledge points, such as "recursive function" and "exception handling", to help them further improve their programming skills; while for a student who is just starting to learn programming, the system will recommend more "easy" level knowledge points, such as "variable assignment" and "conditional statement", to help them lay a good foundation. In summary, this series of steps can not only eliminate the semantic ambiguity of knowledge points and ensure the system's accurate understanding of the content of knowledge points, but also deeply analyze the internal structure and semantic relationship of knowledge points through technologies such as dependency syntax trees and semantic role labeling. On this basis, through semantic network analysis and cluster analysis, the system can evaluate the difficulty of knowledge points and perform reasonable screening, and finally provide students with the most suitable teaching resources for them. For example, on a programming learning platform, the system can use the above method to recommend a series of "medium" difficulty "loop structure" related knowledge points for a student who has a good grasp of "conditional statements" but is not very familiar with "loop structure", including the basic usage and common application scenarios of "for loop" and "while loop", so as to help him gradually overcome learning difficulties and achieve comprehensive improvement of programming skills.

[0038] In a specific embodiment, the structured modeling of the filtered mapped teaching knowledge points to obtain a knowledge system map includes: Performing semantic coding on the filtered mapped teaching knowledge points to obtain semantically coded knowledge points; Constructing a semantic network for the semantically encoded knowledge points to obtain a semantic relationship knowledge point network; The ontology framework is constructed for the semantic relationship knowledge point network by using a preset ontology construction technology to obtain a knowledge point network with an ontology framework; wherein the knowledge point network with an ontology framework includes nodes and edges, wherein the nodes are knowledge points and the edges are relationships between knowledge points; The matching degree between the nodes and the edges is measured by a preset energy function. If the matching degree is not within a preset matching range, the hierarchical structure and relationship constraints of the knowledge points are obtained, and the hierarchical updating of the knowledge point network is performed based on the hierarchical structure and relationship constraints by a preset hierarchical clustering algorithm to obtain a knowledge point network after hierarchical updating; semantically grouping the knowledge point network after the hierarchical update to obtain a semantically grouped knowledge point network; Performing relationship combing on the semantically grouped knowledge point network to obtain a relationship combing knowledge network; Through graph fusion technology, the relationships are organized into a knowledge network for data fusion to obtain a knowledge system graph.

[0039] Specifically, the process of structurally modeling the mapped teaching knowledge points after screening to obtain the knowledge system map is a multi-step, technology-intensive operation. This process aims to organize discrete knowledge points into a structured, hierarchical knowledge network, thereby providing students with a more systematic and intuitive learning path. First, semantic encoding is performed on the mapped teaching knowledge points after screening to obtain semantically encoded knowledge points. Semantic encoding refers to converting knowledge points into a form that can reflect their semantic characteristics, which is usually achieved by mapping the knowledge points to a high-dimensional vector space. For example, in programming learning, the knowledge points "conditional statements" and "loop structures" will be encoded into vector forms, which not only contain the basic information of the knowledge points, but also imply the semantic relationship between them. Through semantic encoding, the system can more accurately understand and process the relationship between knowledge points. Next, a semantic network is constructed for the semantically encoded knowledge points to obtain a semantic relationship knowledge point network. A semantic network is a graph structure in which nodes represent knowledge points and edges represent the relationship between knowledge points. By constructing a semantic network, the system can intuitively display the logical connection and hierarchical structure between knowledge points. For example, the system can use "conditional statements" and "loop structures" as nodes and the logical relationships between them (such as "conditional statements" are often used to control the execution of "loop structures") as edges to construct a preliminary knowledge point network. Then, the semantic relationship knowledge point network is constructed with an ontology framework through the preset ontology construction technology to obtain a knowledge point network with an ontology framework. Ontology construction technology is a method for defining and organizing concepts and their relationships. Through the ontology framework, the system can manage and represent the relationship between knowledge points more systematically. For example, the system can define an ontology framework in which "conditional statements" and "loop structures" belong to the "control structure" category, while "variable assignments" and "function calls" belong to the "basic operations" category. This classification helps the system better organize knowledge points and make their structure clearer. Next, the matching degree between nodes and edges is measured by the preset energy function. If the matching degree is not within the preset matching range, the hierarchical structure and relationship constraints of the knowledge points are obtained, and the knowledge point network is hierarchically updated based on the hierarchical structure and relationship constraints through the preset hierarchical clustering algorithm to obtain the knowledge point network after the hierarchical update. The energy function is used to evaluate the matching degree between nodes and edges. If the matching degree is lower than the preset threshold, the system will readjust the hierarchical structure of the knowledge points. For example, if the system finds that the relationship between "conditional statements" and "loop structures" is not strong enough, it may adjust the "conditional statements" to a more basic level and the "loop structures" to a higher level. Through the hierarchical clustering algorithm, the system can ensure that the hierarchical structure of the knowledge point network is more reasonable and the relationship between knowledge points is closer. Subsequently, the knowledge point network after the hierarchical update is semantically grouped to obtain the semantically grouped knowledge point network.Semantic grouping refers to classifying knowledge points with similar semantic features into the same group, which helps to further optimize the structure of the knowledge point network. For example, the system can group "conditional statements", "loop structures" and "branch structures" into one group because they all belong to the category of control structures. Through semantic grouping, the system can reduce the complexity of the knowledge point network and make it easier to understand and navigate. Then, the semantically grouped knowledge point network is relationally combed to obtain a relationally combed knowledge network. Relationship combing refers to further optimizing the relationship between knowledge points to ensure that the connection between each knowledge point and its related knowledge points is clearer and more accurate. For example, the system can ensure that the relationship between "conditional statements" and "loop structures" is clearly marked as "control structures", while the relationship between "variable assignments" and "function calls" is marked as "data operations". Through relationship combing, the system can improve the logic and coherence of the knowledge point network and make it more in line with students' cognitive patterns. Finally, through graph fusion technology, the relationally combed knowledge network is data fused to obtain a knowledge system map. Graph fusion technology is a method of merging multiple graph structures into a unified graph structure. Through graph fusion, the system can integrate knowledge point networks from different sources into a complete knowledge system map. For example, the system can integrate the programming language basic knowledge network, the data structure knowledge point network and the algorithm knowledge point network into a comprehensive programming knowledge system map. This map can not only show the logical relationship between knowledge points, but also provide students with a systematic learning path to help them better understand and master programming knowledge. In summary, by structurally modeling the filtered mapped teaching knowledge points, the final knowledge system map can not only clearly show the logical relationship and hierarchy between knowledge points, but also provide students with a systematic learning path. For example, on the programming learning platform, the system can build a programming knowledge system map from basic to advanced for students through the above method. In this map, students can start learning from basic knowledge such as "variable assignment" and "data type", gradually transition to control structures such as "conditional statement" and "loop structure", and finally master advanced concepts such as "function call" and "object-oriented programming". In this way, the system can not only help students learn programming knowledge systematically, but also stimulate their learning interest and improve learning effects.

[0040] In a specific embodiment, the recommending teaching resources based on the knowledge system map by using an intelligent decision-making algorithm includes: Based on the knowledge system graph, the learning behavior characteristics are graph matched to obtain a user learning behavior characteristic graph; The similarity between the learning behavior feature graph and the knowledge system graph is calculated by a preset graph similarity calculation algorithm to obtain a similarity matrix; Using a preset graph clustering algorithm, clustering the knowledge system graph based on the similarity matrix to obtain multiple knowledge point clusters; Based on the learning behavior characteristic graph and each of the knowledge point clusters, planning a learning path for the target user to obtain a learning path graph; Using a preset resource recommendation algorithm, based on the learning path map and the learning behavior characteristics, recommend preliminary teaching resources that match the current learning status of the target user; The feedback information of the target users on the recommended preliminary teaching resources is collected through a preset feedback mechanism, and the preliminary teaching resources are dynamically adjusted based on the feedback information to obtain teaching resources.

[0041] Specifically, the process of recommending teaching resources based on the knowledge system map through intelligent decision-making algorithms is a multi-step, highly intelligent operation. This process aims to ensure that the recommended teaching resources not only match the learning behavior characteristics of students, but also can be dynamically adjusted according to the real-time feedback of students, so as to provide a more personalized and effective learning experience. The following is a detailed explanation of this process: First, based on the knowledge system map, the learning behavior characteristics are matched with the map to obtain the user learning behavior characteristic map. The core of this step is to compare the student's personal learning behavior characteristics with the knowledge points in the knowledge system map, so as to generate a map that can reflect the student's current learning status. For example, on a programming learning platform, the system can match this information with the corresponding knowledge points in the knowledge system map by analyzing the student's learning records, such as the videos watched, the exercises completed, and the discussions participated in. Suppose the system finds that a student has mastered "variable assignment" and "conditional statements" but is not familiar with "loop structure", then this information will be mapped to the knowledge system map to generate a map reflecting the student's current learning status. Next, the similarity between the learning behavior feature graph and the knowledge system graph is calculated by the preset graph similarity calculation algorithm to obtain a similarity matrix. The similarity calculation algorithm is used to evaluate the similarity between two graphs, usually by calculating the matching degree between nodes and edges. For example, the system can use the cosine similarity algorithm or other graph matching algorithms to calculate the similarity between each node in the learning behavior feature graph and the corresponding node in the knowledge system graph to generate a similarity matrix. This matrix not only reflects the overall matching degree between the student's current learning status and the knowledge system graph, but also reveals the student's mastery of different knowledge points. Then, the knowledge system graph is clustered based on the similarity matrix using the preset graph clustering algorithm to obtain multiple knowledge point clusters. The graph clustering algorithm is a method for dividing the nodes in the graph into several groups. Through clustering, the system can classify knowledge points with high similarity into the same group, thereby simplifying the structure of the knowledge system graph. For example, the system may classify "variable assignment", "conditional statement" and "loop structure" into one cluster because these knowledge points have a strong correlation in terms of control structure. Through clustering, the system can more clearly display the logical relationship between knowledge points and provide support for subsequent learning path planning. Next, based on the learning behavior characteristic graph and the clustering of each knowledge point, the learning path of the target user is planned to obtain a learning path graph. This step aims to develop a personalized learning plan for students to ensure that they can master the required knowledge points step by step. For example, based on the information in the learning behavior characteristic graph, the system can plan a learning path from "conditional statements" to "loop structures" for the above students, including recommending relevant video tutorials, programming exercises, and case studies. The learning path map not only shows the students' current learning status, but also clarifies the future learning direction and goals.Then, using the preset resource recommendation algorithm, preliminary teaching resources that match the current learning status of the target user are recommended based on the learning path map and learning behavior characteristics. The resource recommendation algorithm usually selects the most suitable learning materials from the resource library in combination with the student's interest preferences, learning habits and current learning progress. For example, the system may recommend a series of video tutorials and programming exercises on "loop structure" to the above students. These resources can not only help students consolidate their existing knowledge, but also guide them to gradually master new knowledge points. In this way, the system can ensure that the recommended teaching resources not only meet the students' current learning needs, but also promote their progress. Finally, through the preset feedback mechanism, the target user's feedback information on the recommended preliminary teaching resources is collected, and the preliminary teaching resources are dynamically adjusted based on the feedback information to obtain the final teaching resources. The feedback mechanism allows students to provide opinions and suggestions in the process of using the recommended resources, and the system will continuously optimize the recommendation results based on these feedback information. For example, if a student expresses difficulty after completing the programming exercise of "loop structure", the system can adjust the recommendation strategy in time and provide more auxiliary resources, such as detailed problem-solving steps and video explanations. Through this dynamic adjustment, the system can better adapt to students' learning changes and provide more personalized learning support. In summary, through intelligent decision-making algorithms, the process of recommending teaching resources based on the knowledge system map can not only accurately capture the current learning status of students, but also dynamically adjust according to students' real-time feedback, thereby providing more personalized and effective learning resources. For example, on a programming learning platform, the system can recommend a series of teaching resources related to "loop structure" to a student who performs well in "conditional statements" but has insufficient mastery of "loop structure" through the above method, and continuously optimize the recommendation results based on students' feedback. This personalized recommendation mechanism can not only improve students' learning efficiency, but also stimulate their interest in learning and promote their all-round development.

[0042] In a specific embodiment, based on the learning behavior characteristic graph and the knowledge point clustering, the learning path of the target user is planned to obtain a learning path graph, including: Step S71, determining the learning path starting point of the target user based on the learning behavior characteristic graph by using a preset path starting point selection algorithm; Step S72, predicting the learning path endpoint of the target user based on the path starting point and the knowledge point clustering by using a preset path endpoint prediction algorithm; Step S73, generating a preliminary learning path based on the path starting point and the path end point; Step S74, based on the preliminary learning path and the knowledge point clustering, optimizing the order of path nodes in the preliminary learning path to obtain an optimized learning path; Step S75, verifying the optimized learning path to determine whether the optimized learning path is reasonable and coherent, and obtaining a verification result; Step S76, if the verification result is qualified, then the optimized learning path is visualized to obtain a learning path diagram; In step S77, if the verification result is not met, the learning behavior characteristic graph is adjusted according to the verification result, and the process returns to step S71 to re-determine the starting point of the learning path of the target user based on the adjusted learning behavior characteristic graph until the optimized learning path is reasonable and coherent.

[0043] Specifically, in the process of planning the learning path of the target user based on the learning behavior characteristic graph and knowledge point clustering, the entire process can be carefully divided into seven consecutive steps, each of which is to ensure that the learning path graph finally generated can accurately reflect the learning needs of the target user, while ensuring the rationality and coherence of the learning path. The specific implementation method of each step and the logic behind it will be elaborated in detail in the order of the above steps. First, step S71 determines the starting point of the learning path of the target user based on the learning behavior characteristic graph through a preset path starting point selection algorithm. This means that the system needs to analyze the user's learning behavior characteristic graph and identify the node that best represents the user's current knowledge level and learning interest as the starting point of the learning path. For example, in the application scenario of programming learning, the system may analyze the user's online learning records and find that the user has recently frequently viewed the content related to "function definition", but the understanding of "function call" is not deep enough. Therefore, the system will select "function call" as the starting point of the learning path, so that the user can start from the knowledge points he is familiar with and gradually transition to more complex concepts. Then, step S72 predicts the end point of the learning path of the target user based on the path starting point and knowledge point clustering through a preset path end point prediction algorithm. The key to this step is to use algorithms to predict the knowledge level and skill level that users should reach after completing a certain stage of learning. The system will comprehensively consider the user's starting point position and other related knowledge points in the knowledge point cluster to infer the user's learning end point. For example, for users who choose "function call" as the starting point, the system may predict that their learning end point is "object-oriented programming", because this is an important direction to further deepen programming capabilities after mastering basic function operations. Subsequently, step S73 generates a preliminary learning path based on the path starting point and path end point. The task of this step is to build a learning path connecting the two points based on the determination of the starting point and end point. This path will contain a series of ordered knowledge points, aiming to guide users to smoothly transition from their existing knowledge level to the expected target level. Continuing with programming learning as an example, the path from "function call" to "object-oriented programming" may include knowledge points such as "function parameter passing", "scope and life cycle", and "class and object". These knowledge points are arranged in a certain logical order to ensure that users can gradually build a complete knowledge system. Next, step S74 optimizes the order of path nodes in the preliminary learning path based on the preliminary learning path and knowledge point clustering to obtain an optimized learning path. The purpose of optimization is to ensure that the learning path not only covers the necessary knowledge points, but also that the connection between these knowledge points is as natural and smooth as possible, which helps users absorb new knowledge efficiently. For example, the system may arrange "scope and life cycle" before "function parameter passing" based on the dependencies between knowledge points, because understanding the scope of variables is a prerequisite for the correct use of parameters.In addition, the system can also consider the principle of increasing difficulty of knowledge points to avoid suddenly increasing the difficulty of learning and affecting the user experience. After that, step S75 verifies the optimized learning path to determine whether the optimized learning path is reasonable and coherent, and obtains the verification result. The verification process may involve a variety of methods such as expert review and historical data comparison, with the aim of ensuring that the design of the learning path is both scientific and practical. If it is found that there are unreasonable aspects in the path design, such as too large jumps between certain knowledge points, or a lack of sufficient transitional knowledge points, the system will record these problems to provide a basis for subsequent adjustments. If the verification results show that the optimized learning path is reasonable and coherent, step S76 visualizes the optimized learning path to obtain a learning path map. The learning path map will be presented to users in a graphical way to help them intuitively understand their learning route and clarify their learning goals. Users can easily find their current position and what they should learn next through the learning path map, which is crucial to maintaining learning motivation and sense of direction. However, if the verification results show that the optimized learning path is not reasonable and coherent, step S77 will adjust the learning behavior characteristic diagram according to the verification results, and return to step S71 to re-determine the starting point of the target user's learning path based on the adjusted learning behavior characteristic diagram until the optimized learning path is reasonable and coherent. This means that the entire planning process is an iterative optimization process. Only when the system confirms that the learning path fully meets the user's needs and is reasonably designed will it stop adjusting and finally determine the learning path. In short, through such a carefully designed process, the system can not only provide users with personalized and targeted learning paths, but also continuously optimize the path design throughout the learning process to ensure the maximization of learning effects. For example, in the programming learning platform, the system plans a learning path from basic to advanced for a user who is lacking in "function calls" through the above process, which not only helps users fill in knowledge gaps, but also promotes the overall improvement of their programming skills.

[0044] The above describes the method for recommending artificial intelligence teaching resources for primary and secondary schools in the embodiment of the present invention. The following describes the system for recommending artificial intelligence teaching resources for primary and secondary schools in the embodiment of the present invention. Figure 2 In one embodiment of the present invention, a recommendation system for primary and secondary school artificial intelligence teaching resources includes: An analysis module 21 is used to analyze the learning behavior of target users through educational data mining technology to obtain learning behavior characteristics; A mapping module 22, used to use a preset graph embedding algorithm to perform association mapping on knowledge points in preset educational resources based on the learning behavior characteristics to obtain mapped teaching knowledge points; A screening module 23 is used to analyze the difficulty of the mapping teaching knowledge points through a preset natural language processing technology, and screen the mapping teaching knowledge points based on the difficulty to obtain screened mapping teaching knowledge points; A modeling module 24 is used to perform structured modeling on the filtered mapped teaching knowledge points to obtain a knowledge system map; The recommendation module 25 is used to recommend teaching resources based on the knowledge system map through an intelligent decision-making algorithm.

[0045] In this embodiment, for the specific implementation of each unit in the above system embodiment, please refer to the above method embodiment, which will not be repeated here.

[0046] Reference Figure 3 The present invention also provides a computer device in an embodiment, wherein the internal structure of the computer device can be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0047] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0048] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0049] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM.

[0050] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0051] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for recommending artificial intelligence teaching resources for primary and secondary schools, characterized in that: The following steps are involved: Analyze the learning behavior of target users through educational data mining technology to obtain learning behavior characteristics; Using the preset TransE algorithm, based on the learning behavior characteristics, the knowledge points in the preset educational resources are associated and mapped to obtain mapped teaching knowledge points; Analyzing the difficulty of the mapped teaching knowledge points through a preset natural language processing technology, and screening the mapped teaching knowledge points based on the difficulty to obtain screened mapped teaching knowledge points; Performing structured modeling on the filtered mapped teaching knowledge points to obtain a knowledge system map; Through an intelligent decision-making algorithm, teaching resources are recommended based on the knowledge system map.

2. The method for recommending artificial intelligence teaching resources for primary and secondary schools according to claim 1, characterized in that: The learning behavior analysis of the target users is performed by using educational data mining technology to obtain learning behavior characteristics, including: By using a preset statistical analysis method, historical learning data of the target user is analyzed to obtain a learning activity curve; wherein the learning activity curve includes the learning activity and learning time distribution of the target user, and the learning activity curve is a curve with the learning activity as the Y-axis and the learning time distribution as the X-axis; Classifying the learning behavior pattern of the target user based on the learning activity curve to obtain a learning feature pattern; Based on the learning feature pattern, the target user's emotion change is identified to obtain an emotion change trend; wherein the emotion change trend is the target user's emotion state at different learning stages, and the emotion change trend includes a positive state, a neutral state, and a negative state; Through educational data mining technology, historical learning data mining is performed on the target user based on the emotion change trend to obtain learning behavior characteristics; wherein the learning behavior characteristics include interest preferences and learning habits.

3. The method for recommending artificial intelligence teaching resources for primary and secondary schools according to claim 1, characterized in that: The preset TransE algorithm is used to associate and map the knowledge points in the preset educational resources based on the learning behavior characteristics to obtain the mapped teaching knowledge points, including: Using a semantic network construction algorithm, the learning behavior characteristics are converted into semantic network structure data; Based on a word vector algorithm, vector mapping is performed on the learning behavior features in the semantic network to obtain a mapping feature vector; Determining the weight of the mapping feature vector according to the importance of the learning behavior feature through a vector weighting algorithm to obtain a learning behavior weight vector; Converting the knowledge points in the educational resource into vector format to obtain knowledge point vectors; By using a preset cosine similarity algorithm, the angle between the knowledge point vector and the learning behavior weight vector in the vector space is calculated to obtain a cosine similarity value between the knowledge point vector and the learning behavior weight vector; Sorting the knowledge points corresponding to the knowledge point vectors according to the cosine similarity values ​​to obtain a sorted knowledge point sequence; The preset TransE algorithm is used to map the sorted knowledge point sequence into a low-dimensional vector space to obtain mapped teaching knowledge points.

4. The method for recommending artificial intelligence teaching resources for primary and secondary schools according to claim 1, characterized in that: The method of analyzing the difficulty of the mapped teaching knowledge points by using a preset natural language processing technology, and screening the mapped teaching knowledge points based on the difficulty to obtain the screened mapped teaching knowledge points includes: By using a preset semantic ambiguity model, it is determined whether the mapping teaching knowledge point has semantic ambiguity. If so, the mapping teaching knowledge point is semantically clarified by using a preset semantic disambiguation technology to obtain semantically clear knowledge points; Constructing a dependency syntax tree for the semantically clear knowledge point to obtain a knowledge point with a dependency syntax tree; By using the semantic role labeling module in the semantic ambiguity model, the knowledge points with the dependency syntax tree are labeled with semantic roles to obtain the knowledge points labeled with semantic roles; Using a preset semantic network analysis algorithm, the semantic importance of the knowledge points labeled with the semantic roles is evaluated to obtain a knowledge point importance vector; Analyzing the degree of difficulty of the knowledge point importance vector by the cluster analysis module in the semantic network analysis algorithm, and clustering the knowledge point importance vector based on the degree of difficulty to obtain a cluster analysis result; wherein the cluster analysis result includes a vector with difficulty clustering; The mapped teaching knowledge points are screened based on the cluster analysis results through a preset natural language processing technology to obtain screened mapped teaching knowledge points.

5. The method for recommending artificial intelligence teaching resources for primary and secondary schools according to claim 1, characterized in that: The structured modeling of the filtered mapped teaching knowledge points to obtain a knowledge system map includes: Performing semantic coding on the filtered mapped teaching knowledge points to obtain semantically coded knowledge points; Constructing a semantic network for the semantically encoded knowledge points to obtain a semantic relationship knowledge point network; The ontology framework is constructed for the semantic relationship knowledge point network by using a preset ontology construction technology to obtain a knowledge point network with an ontology framework; wherein the knowledge point network with an ontology framework includes nodes and edges, wherein the nodes are knowledge points and the edges are relationships between knowledge points; The matching degree between the nodes and the edges is measured by a preset energy function. If the matching degree is not within a preset matching range, the hierarchical structure and relationship constraints of the knowledge points are obtained, and the hierarchical updating of the knowledge point network is performed based on the hierarchical structure and relationship constraints by a preset hierarchical clustering algorithm to obtain a knowledge point network after hierarchical updating; semantically grouping the knowledge point network after the hierarchical update to obtain a semantically grouped knowledge point network; Performing relationship combing on the semantically grouped knowledge point network to obtain a relationship combing knowledge network; Through graph fusion technology, the relationships are organized into a knowledge network for data fusion to obtain a knowledge system graph.

6. The method for recommending artificial intelligence teaching resources for primary and secondary schools according to claim 1, characterized in that: The method of recommending teaching resources based on the knowledge system map by using an intelligent decision-making algorithm includes: Based on the knowledge system graph, the learning behavior characteristics are graph matched to obtain a user learning behavior characteristic graph; The similarity between the learning behavior feature graph and the knowledge system graph is calculated by a preset graph similarity calculation algorithm to obtain a similarity matrix; Using a preset graph clustering algorithm, clustering the knowledge system graph based on the similarity matrix to obtain multiple knowledge point clusters; Based on the learning behavior characteristic graph and each of the knowledge point clusters, planning a learning path for the target user to obtain a learning path graph; Using a preset resource recommendation algorithm, based on the learning path map and the learning behavior characteristics, recommend preliminary teaching resources that match the current learning status of the target user; The feedback information of the target users on the recommended preliminary teaching resources is collected through a preset feedback mechanism, and the preliminary teaching resources are dynamically adjusted based on the feedback information to obtain teaching resources.

7. The method for recommending artificial intelligence teaching resources for primary and secondary schools according to claim 6, characterized in that: Based on the learning behavior characteristic graph and the knowledge point clustering, the learning path of the target user is planned to obtain a learning path graph, including: Step S71, determining the learning path starting point of the target user based on the learning behavior characteristic graph by using a preset path starting point selection algorithm; Step S72, predicting the learning path endpoint of the target user based on the path starting point and the knowledge point clustering by using a preset path endpoint prediction algorithm; Step S73, generating a preliminary learning path based on the path starting point and the path end point; Step S74, based on the preliminary learning path and the knowledge point clustering, optimizing the order of path nodes in the preliminary learning path to obtain an optimized learning path; Step S75, verifying the optimized learning path to determine whether the optimized learning path is reasonable and coherent, and obtaining a verification result; Step S76, if the verification result is qualified, then the optimized learning path is visualized to obtain a learning path diagram; In step S77, if the verification result is not met, the learning behavior characteristic graph is adjusted according to the verification result, and the process returns to step S71 to re-determine the starting point of the learning path of the target user based on the adjusted learning behavior characteristic graph until the optimized learning path is reasonable and coherent.

8. A recommendation system for artificial intelligence teaching resources in primary and secondary schools, characterized in that: include: The analysis module is used to analyze the learning behavior of target users through educational data mining technology to obtain learning behavior characteristics; A mapping module, used to use a preset graph embedding algorithm to perform association mapping on the knowledge points in the preset educational resources based on the learning behavior characteristics to obtain mapped teaching knowledge points; A screening module, used to analyze the difficulty of the mapping teaching knowledge points through a preset natural language processing technology, and screen the mapping teaching knowledge points based on the difficulty to obtain screened mapping teaching knowledge points; A modeling module, used to perform structured modeling on the filtered mapped teaching knowledge points to obtain a knowledge system map; The recommendation module is used to recommend teaching resources based on the knowledge system map through an intelligent decision-making algorithm.

9. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images