Learning path intelligent recommendation system based on user behavior big data analysis

Abstract

The present invention relates to the technical field of learning path recommendation, and discloses an intelligent learning path recommendation system based on user behavior big data analysis. The system comprises: a behavior data collection module for collecting learners' behavior data on a learning platform; a behavior inertia modeling module for extracting learners' behavior inertia parameters based on the behavior data; a task capability map construction module for constructing a task capability map; calculating a mastery score based on the test results and extracting a subgraph map; and a recommended path generation module for generating an optimal learning recommendation path based on a path recommendation engine, and selecting the optimal learning recommendation path as a recommendation result based on the path fitness score. The present invention realizes the dynamic generation of personalized, rhythm-adapted, and resource-optimized learning path recommendations based on user behavior big data analysis and combined with job capability requirements.

Description

Technical Field

[0001] The present invention belongs to the technical field of learning path recommendation, and specifically relates to an intelligent learning path recommendation system based on big data analysis of user behavior. Background Art

[0002] With the rapid development of vocational education, vocational colleges are increasingly focusing on providing personalized learning path design based on different professional directions and job competency requirements. Traditional teaching models often adopt a unified curriculum schedule and a fixed teaching schedule, which makes it difficult to fully consider the differences among learners in terms of learning rhythm, cognitive ability, knowledge mastery, and resource usage preferences. This leads to unsatisfactory learning results and makes it difficult to accurately achieve job competency training goals. Some existing path recommendation methods are mainly based on simple sorting of course plans or basic prerequisites. They lack in-depth mining and dynamic analysis of learners' learning behavior data, and are unable to adjust recommended paths in a timely manner according to learners' actual learning status. At the same time, current intelligent applications in vocational education focus more on resource integration and information management, and rarely implement dynamic modeling and application of systematic relationships between job tasks, competency indicators, and course knowledge points. Summary of the Invention

[0003] The present invention provides an intelligent learning path recommendation system based on big data analysis of user behavior, which solves the technical problems in related technologies such as the inability to dynamically adjust learning paths according to learners' actual behavior and the lack of job ability orientation and personalized rhythm matching.

[0004] The present invention provides an intelligent learning path recommendation system based on big data analysis of user behavior, including:

[0005] The behavioral data collection module is used to collect learners' behavioral data on the learning platform, including click behavior, learning time, learning completion rate and test scores;

[0006] A behavioral inertia modeling module is used to extract behavioral inertia parameters of learners based on the behavioral data, wherein the behavioral inertia parameters include: learning rhythm stability coefficient, cognitive load fluctuation coefficient, and resource type effectiveness characteristics;

[0007] A task capability map construction module is used to construct a task capability map based on the association between job tasks and capability indicators, capability indicators and course modules, and course modules and knowledge points predefined on the learning platform. The task capability map includes job task nodes, capability indicator nodes, course module nodes, and knowledge point nodes, as well as the associated edges between the nodes;

[0008] Calculate the mastery score of each capability indicator node based on the test results;

[0009] Extract sub-graphs from the task ability graph based on mastery scores and behavioral inertia parameters;

[0010] The recommended path generation module is used to generate the optimal learning recommendation path based on the behavioral inertia parameters and the subgraph map according to the path recommendation engine. The path recommendation engine is used to generate Q candidate learning recommendation paths and select the optimal learning recommendation path as the recommendation result based on the path fitness score. The candidate learning recommendation path is a sequence of course module nodes and knowledge point nodes arranged in sequence, and Q is a positive integer.

[0011] Furthermore, a temporal rhythm dispersion coefficient is constructed by calculating the ratio of the standard deviation of the learning peak time point to the mean of the learning peak time point. A first intermediate feature is obtained by calculating the Pearson coefficient between the learner's learning time and test score in a first preset time period and normalizing it using a sigmoid function. The temporal rhythm dispersion coefficient and the first intermediate feature are combined to obtain a learning rhythm stability coefficient.

[0012] The error rate deviation coefficient is constructed based on the ratio of the difference between the number of wrong questions in the learner's test and the mean number of wrong questions in the test and the standard deviation of the number of wrong questions in the test. The behavioral operation abnormality coefficient is constructed based on the average of the learner's page dwell time, the number of page return times in a single learning session, and the number of page return times in a single learning session of other learners on the platform. The error rate deviation coefficient and the operation abnormality coefficient are combined to obtain the cognitive load fluctuation coefficient.

[0013] The second intermediate feature is obtained by the ratio of the learner's learning time on one resource type to the total learning time on all resource types. The third intermediate feature is obtained by the ratio of the learner's average test score to the full score of the test. The second and third intermediate features are combined to obtain the resource type effectiveness feature.

[0014] Furthermore, the relationships in the task capability graph include:

[0015] There is a one-to-many relationship between job task nodes and capability indicator nodes;

[0016] There is a one-to-many association between capability indicator nodes and course module nodes;

[0017] There is a one-to-many association between course module nodes and knowledge point nodes;

[0018] Directed edges are established between knowledge point nodes based on the previous and next dependencies of the knowledge system;

[0019] Directed edges are generated between nodes of different types to construct a nested graph structure of job tasks-competency indicators-course modules-knowledge points.

[0020] Furthermore, the mastery score of the capability indicator node is calculated based on the test scores associated with the capability indicator node, wherein the test scores of the capability indicator node are normalized and weighted summed to obtain the mastery score.

[0021] Furthermore, based on the mastery score and behavioral inertia parameters, a subgraph is extracted from the task capability graph. The specific steps include:

[0022] S201, screening ability indicator nodes whose mastery score is greater than a first preset threshold and whose learning rhythm stability coefficient is greater than a second preset threshold;

[0023] S202, extracting, based on the screened capability indicator nodes, the job task nodes and course module nodes connected thereto, as well as the knowledge point nodes connected to the course module nodes;

[0024] S203, obtaining a subgraph map based on the screened capability indicator nodes and the extracted job task nodes, course module nodes and knowledge point nodes, as well as the edges between the nodes.

[0025] Furthermore, the recommended path generation module imposes the following internal constraints during the candidate learning recommended path generation process executed by the path recommendation engine:

[0026] The order of course module nodes and knowledge point nodes is the learning order between course module nodes and knowledge point nodes in the subgraph map;

[0027] There are edges between adjacent knowledge point nodes in the learner's candidate learning recommendation path;

[0028] The estimated total learning time of the learner's candidate recommended learning paths does not exceed a fourth preset threshold.

[0029] Furthermore, the recommended path generation module further includes: a module for receiving intervention information input from an external port as an external constraint imposed by the path recommendation engine in the process of generating candidate learning recommended paths, wherein the intervention information includes:

[0030] Specify the mandatory priority learning order of course module nodes;

[0031] Specify that the candidate learning recommendation path contains specific knowledge point nodes;

[0032] The estimated total learning time of the designated candidate learning recommendation path is not less than a fifth preset threshold.

[0033] Furthermore, the recommended path generation module performs structured coding on the candidate learning recommended paths, and the specific steps include:

[0034] S301: hierarchically encode the course module nodes and knowledge point nodes in the candidate learning recommendation path according to their dependency relationships in the subgraph to generate a node sequence with a topological sequence number;

[0035] S302 , mapping the resource type efficiency feature in the behavioral inertia parameter into a node weight label, and forming a coding vector with the node sequence generated in S301 .

[0036] Furthermore, the calculation process of the path fitness score includes:

[0037] S401: Detect nodes in the node sequence that violate the dependency relationship in the subgraph graph as conflict nodes, assign penalty values ​​to the conflict nodes, and generate a sequence rationality score based on the hierarchical depth of the conflict nodes;

[0038] S402, setting the difficulty level of course module nodes and knowledge point nodes, counting the node pairs in the candidate learning recommendation path whose difficulty level difference exceeds 2 levels, and obtaining the difficulty gradient smoothness score through exponential decay calculation;

[0039] S403 , performing weighted addition on the sequence rationality score and the difficulty gradient smoothness score to obtain a path fitness score.

[0040] Furthermore, the step of selecting the optimal learning recommendation path based on the path fitness score by the recommended path generation module includes:

[0041] S501, sorting the Q candidate learning recommendation paths in descending order according to the path fitness scores, and retaining a preset number of candidate learning recommendation paths;

[0042] S502: Perform a first mutation operation on the retained candidate learning recommendation path with a first preset probability. The first mutation operation includes:

[0043] Randomly select a course module node in the candidate learning recommendation path;

[0044] Retrieve knowledge point nodes associated with the course module nodes but not in the candidate learning recommendation path from the subgraph map, and filter out knowledge point nodes whose resource type effectiveness characteristics are higher than a sixth preset threshold;

[0045] Replace the knowledge point nodes connected to the course module nodes in the node sequence with the filtered knowledge point nodes;

[0046] S503, recalculating the path fitness score for the candidate learning recommendation path after the first mutation operation;

[0047] S504, repeatedly execute S501 to S503 until the improvement of the highest path fitness score among the candidate learning paths in consecutive U rounds of iterations does not exceed the seventh preset threshold, and the candidate learning recommendation path with the highest path fitness score in the U round is used as the optimal learning recommendation path, where U is a positive integer.

[0048] The beneficial effects of the present invention are as follows: by collecting multi-dimensional behavioral data of learners on the learning platform, extracting behavioral inertia parameters, combining job tasks, ability indicators, course modules and knowledge points to construct a multi-layer nested task ability map, dynamically extracting personalized subgraphs, and based on the path fitness score and variation optimization mechanism, generating the optimal learning recommendation path that meets the learner's personality characteristics and job requirements; by introducing external intervention constraints, achieving coordinated regulation of path recommendation with teaching objectives and job requirements. The present invention can effectively improve the personalized matching degree of learning paths, optimize learning rhythm and cognitive load distribution, improve resource utilization efficiency, and enhance the job competence and learning completion quality of vocational school students. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a module diagram of the learning path intelligent recommendation system based on user behavior big data analysis of the present invention. DETAILED DESCRIPTION

[0050] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed solely to enable those skilled in the art to better understand and implement the subject matter described herein, and that the functions and arrangements of the elements discussed may be varied without departing from the scope of this specification. Various examples may omit, substitute, or add various processes or components as needed. In addition, features described with respect to some examples may also be combined in other examples.

[0051] like Figure 1 As shown in the figure, the learning path intelligent recommendation system based on user behavior big data analysis includes:

[0052] The behavior data collection module 101 is used to collect the behavior data of learners on the learning platform, wherein the behavior data includes: click behavior, learning time, learning completion rate and test scores;

[0053] A behavior inertia modeling module 102 is configured to extract behavior inertia parameters of the learner based on the behavior data, wherein the behavior inertia parameters include: a learning rhythm stability coefficient, a cognitive load fluctuation coefficient, and a resource type effectiveness characteristic;

[0054] A task capability map construction module 103 is used to construct a task capability map based on the association relationships between job tasks and capability indicators, capability indicators and course modules, and course modules and knowledge points predefined on the learning platform. The task capability map includes job task nodes, capability indicator nodes, course module nodes, and knowledge point nodes, as well as the associated edges between the nodes.

[0055] Calculate the mastery score of each capability indicator node based on the test results;

[0056] Extract sub-graphs from the task ability graph based on mastery scores and behavioral inertia parameters;

[0057] The recommended path generation module 104 is used to generate the optimal learning recommendation path based on the behavioral inertia parameters and the subgraph map according to the path recommendation engine, wherein the path recommendation engine is used to generate Q candidate learning recommendation paths, and select the optimal learning recommendation path as the recommendation result based on the path fitness score. The candidate learning recommendation path is a sequence of a set of sequentially arranged course module nodes and knowledge point nodes, and Q is a positive integer.

[0058] In one embodiment of the present invention, by embedding points on the front-end page, click behaviors are obtained and recorded, and click behaviors include: resource access clicks, page jump clicks, control operation clicks and page back operations; the learning time is obtained by recording the time when the learner enters and leaves the page; the learning completion rate is calculated by reading the video playback progress, the number of completed test questions, etc.; the learner's test scores are obtained through the back-end interface; the above data have different sources and inconsistent dimensions, so in order to ensure the accuracy of subsequent data fusion and calculations, they are normalized.

[0059] In one embodiment of the present invention, the calculation formula of the learning rhythm stability coefficient is: ,in, Indicates the learning rhythm stability coefficient, which is used to measure the overall rhythm stability. The higher the learning rhythm stability, the more stable it is. represents the standard deviation of the learning peak time point, which is used to measure volatility. represents the mean of the learning peak time point, represents the regular learning time point, sigmoid represents the sigmoid function, Represents a sequence of N days of study time, Represents a sequence of the corresponding test scores for the next day. express and The Pearson correlation coefficient between the two is used to measure the consistency between learning behavior and learning effect;

[0060] The calculation formula of the cognitive load fluctuation coefficient is: ,in, It represents the cognitive load fluctuation coefficient, which measures the cognitive fluctuation degree of learners in the learning process by analyzing the fluctuation degree of error performance and the abnormal intensity of behavioral operations. represents the tanh function, represents the number of wrong questions in the t-th test, It represents the average number of wrong questions in the last 10 tests. Indicates the standard deviation of the number of wrong questions in the last 10 tests. Indicates the number of page rollbacks in a single learning session, d indicates the time spent on the current page, Indicates the average residence time of similar resources. Indicates the average number of page rollbacks during a single learning session for other learners on the platform;

[0061] The calculation formula for the resource type efficiency characteristics is: ,in, Represents the resource type effectiveness feature, which is used to measure the learning promotion effect of a resource type on the current user. represents the learning time of the learner on resource type r, Indicates the learning time of learners on all resource types. Specifically, resource types include: course videos, course documents, course examination questions and external link resources. represents the average of the learners’ test scores, Indicates the full score of the test.

[0062] In one embodiment of the present invention, the task capability map is used to represent the multi-level association relationship between job tasks, capability indicators, course modules and knowledge points in the learning platform;

[0063] Among them, the task capability map includes: job task nodes, capability indicator nodes, course module nodes and knowledge point nodes; job task nodes represent the task objectives that need to be completed under the actual job; capability indicator nodes: represent the specific capability items required to complete job tasks; course module nodes represent the modular teaching units for capability training in the course system provided by the platform; knowledge point nodes represent the specific knowledge content units subdivided in each course module.

[0064] In one embodiment of the present invention, the association relationships in the task capability map include:

[0065] There is a one-to-many relationship between job task nodes and capability indicator nodes;

[0066] There is a one-to-many association between capability indicator nodes and course module nodes;

[0067] There is a one-to-many association between course module nodes and knowledge point nodes;

[0068] Directed edges are established between knowledge point nodes based on the previous and next dependencies of the knowledge system;

[0069] Directed edges are generated between nodes of different types to construct a nested graph structure of job tasks-competency indicators-course modules-knowledge points. The directed edges between nodes point from job task nodes to competency indicator nodes, from competency indicator nodes to course module nodes, and from course module nodes to knowledge point nodes. The directed edges between knowledge points are determined according to the order in which knowledge is mastered.

[0070] By constructing a nested graph structure of job tasks-competency indicators-course modules-knowledge points, we can achieve full-chain mapping from job requirements to knowledge point mastery paths, making learning recommendations more accurate and systematic.

[0071] In one embodiment of the present invention, the mastery score of the capability indicator node is calculated based on the test scores associated with the capability indicator node, wherein the test scores of the capability indicator node are normalized and weighted summed to obtain the mastery score. The calculation formula of the mastery score is: , represents the mastery of the capability indicator node c, c represents the index of the capability indicator node, M represents the number of tests, j represents the index of the test, represents the learner's test score at the jth ability indicator node c, represents the full score of the j-th test.

[0072] In one embodiment of the present invention, a subgraph is extracted from a task capability graph based on a mastery score and a behavioral inertia parameter. Specifically, the steps include:

[0073] S201, screening ability indicator nodes whose mastery score is greater than a first preset threshold and whose learning rhythm stability coefficient is greater than a second preset threshold;

[0074] S202, extracting, based on the screened capability indicator nodes, the job task nodes and course module nodes connected thereto, as well as the knowledge point nodes connected to the course module nodes;

[0075] S203, obtaining a subgraph map based on the screened capability indicator nodes and the extracted job task nodes, course module nodes and knowledge point nodes, as well as the edges between the nodes.

[0076] In one embodiment of the present invention, the recommended path generation module imposes the following internal constraints during the candidate learning recommended path generation process executed by the path recommendation engine:

[0077] The order of course module nodes and knowledge point nodes is the learning order between course module nodes and knowledge point nodes in the subgraph;

[0078] There are edges between adjacent knowledge point nodes in the learner's candidate learning recommendation path to ensure the coherence of the knowledge points in the learning path;

[0079] The estimated total learning time of the learner's candidate learning recommendation path does not exceed a fourth preset threshold, wherein the estimated total learning time is estimated based on the learner's learning time in the historical learning process.

[0080] In one embodiment of the present invention, the recommended path generation module further includes: receiving intervention information input from external ports as external constraints imposed by the path recommendation engine during the generation of candidate learning recommendation paths. The external ports include: a teacher management terminal, an enterprise learning management platform, and a third-party teaching interface. The intervention information includes:

[0081] Specify the mandatory priority learning order of course module nodes to ensure that key courses are presented first;

[0082] Specify specific knowledge points in the recommended candidate learning paths to prevent the learning paths from missing core knowledge content of the position;

[0083] The estimated total learning time of the candidate learning recommendation path is specified to be no less than the fifth preset threshold to avoid the path length being too short, resulting in insufficient learning depth.

[0084] In one embodiment of the present invention, the step of generating Q candidate learning recommendation paths by the path recommendation engine includes:

[0085] According to the sequential dependency relationship between the course module nodes and the knowledge point nodes in the subgraph, the subgraph is traversed to construct a node sequence that meets the internal and external constraints.

[0086] Randomly insert node switching and backtracking operations during the generation process to avoid excessive concentration of all paths at the beginning or end;

[0087] When the number of candidate learning recommendation paths generated reaches Q, the generation is stopped.

[0088] In one embodiment of the present invention, the recommended path generation module performs structured coding on the candidate learning recommended paths, and the specific steps include:

[0089] S301: Hierarchically encode the course module nodes and knowledge point nodes in the candidate learning recommendation path according to their dependency relationships in the subgraph map to generate a node sequence with a topological sequence number. Specifically, the topological sequence number represents the sequence number of the node, and the node sequence is expressed as: ;

[0090] S302: Map the resource type efficiency features in the behavioral inertia parameters into node weight labels, and form an encoding vector with the node sequence generated in S301. Specifically, the node weight label is a characteristic value assigned to the resource type of each node in the candidate learning recommendation path. For example, the node sequence is: , the encoding vector is:

[0091]

[0092] In one embodiment of the present invention, the calculation process of the path fitness score includes:

[0093] S401: Detect nodes in the node sequence that violate the dependency relationship in the subgraph graph as conflict nodes, assign penalty values ​​to the conflict nodes, and generate a sequence rationality score based on the hierarchical depth of the conflict nodes. The calculation formula for the sequence rationality score is: ,in, It represents the order rationality score, which is used to measure the degree to which the arrangement order of course module nodes and knowledge point nodes in the candidate learning recommendation path complies with the knowledge system dependency and teaching logic, ensuring the coherence of the learning process and the rationality of the knowledge structure. P represents the number of conflicting nodes, and o represents the index of the conflicting node. Indicates the level depth of the conflict node in the subgraph. For example, the level depth of the course module node is 1, and the level depth of the knowledge point node is 2. represents the first weight coefficient;

[0094] S402: Set the difficulty level of the course module nodes and knowledge point nodes, count the node pairs in the candidate learning recommendation path whose difficulty level difference exceeds 2 levels, and calculate the difficulty gradient smoothness score through exponential decay. The calculation formula of the difficulty gradient smoothness score is: ,in, represents the difficulty gradient smoothness score, which is used to evaluate the smoothness of the change in difficulty level of adjacent content in the candidate learning recommendation path. e represents a natural constant. represents the attenuation intensity coefficient, represents the tolerance factor, Indicates the number of node pairs whose difficulty levels differ by more than 2 levels;

[0095] S403: Perform weighted addition of the sequence rationality score and the difficulty gradient smoothness score to obtain a path fitness score. The calculation formula for the path fitness score is: ,in, It represents the path fitness score, which is used to comprehensively evaluate the overall optimization degree of the candidate learning recommendation path in terms of knowledge coherence and difficulty transition smoothness, quantify the matching level between the path and user behavior characteristics and knowledge system requirements, and guide the present invention to screen out the optimal learning recommendation path. and represent the second weight coefficient and the third weight coefficient respectively.

[0096] In one embodiment of the present invention, the step of selecting the optimal learning recommendation path based on the path fitness score by the recommended path generation module includes:

[0097] S501, sorting the Q candidate learning recommendation paths in descending order according to the path fitness scores, and retaining a preset number of candidate learning recommendation paths;

[0098] S502: Perform a first mutation operation on the retained candidate learning recommendation path with a first preset probability. The first mutation operation includes:

[0099] Randomly select a course module node in the candidate learning recommendation path;

[0100] Retrieve knowledge point nodes associated with the course module nodes but not in the candidate learning recommendation path from the subgraph map, and screen out knowledge point nodes whose resource type effectiveness characteristics are higher than a sixth preset threshold;

[0101] Replace the knowledge point nodes connected to the course module nodes in the node sequence with the filtered knowledge point nodes;

[0102] S503, recalculating the path fitness score for the candidate learning recommendation path after the first mutation operation;

[0103] S504, repeatedly execute S501 to S503 until the improvement of the highest path fitness score among the candidate learning paths in consecutive U rounds of iterations does not exceed the seventh preset threshold, and the candidate learning recommendation path with the highest path fitness score in the U round is used as the optimal learning recommendation path, where U is a positive integer.

[0104] Through screening based on path fitness scores and dynamic optimization of the first mutation operation, it is possible to maintain the rationality of the recommended learning path while guiding learners to prioritize knowledge units with high resource efficiency and excellent learning experience, avoid path recommendations falling into local optimality, improve the personalized adaptability of the recommended learning path and the quality of learning completion, and enhance learning efficiency and job competency achievement.

[0105] It should be noted that the intervals and thresholds are set for ease of comparison. The threshold size depends on the amount of sample data and the cardinality set by those skilled in the art for each set of sample data, as long as it does not affect the proportional relationship between the parameter and the quantized value. Furthermore, the above formulas are all dimensionless numerical calculations. These formulas are derived from software simulations of the most recent real-world conditions using large amounts of data. The preset parameters in these formulas are set by those skilled in the art based on actual conditions.

[0106] The above describes the embodiments of the present invention, but the present invention is not limited to the above specific implementation methods. The above specific implementation methods are merely illustrative and not restrictive. Ordinary technicians in this field can also make many forms based on the inspiration of this embodiment, all of which are protected by this embodiment.

Claims

1. An intelligent learning path recommendation system based on big data analysis of user behavior, characterized by: include: The behavioral data collection module is used to collect learners' behavioral data on the learning platform, including click behavior, learning time, learning completion rate and test scores; A behavioral inertia modeling module is used to extract behavioral inertia parameters of learners based on the behavioral data, wherein the behavioral inertia parameters include: learning rhythm stability coefficient, cognitive load fluctuation coefficient, and resource type effectiveness characteristics; The calculation formula of the learning rhythm stability coefficient is: , represents the learning rhythm stability coefficient, represents the standard deviation of the learning peak time point, Represents the mean of the learning peak time point, sigmoid represents the sigmoid function, Represents a sequence of N days of study time, Represents a sequence of the corresponding test scores for the next day. express and Pearson correlation coefficient between ; The calculation formula of the cognitive load fluctuation coefficient is: ,in, represents the cognitive load fluctuation coefficient, represents the tanh function, represents the number of wrong questions in the t-th test, It represents the average number of wrong questions in the last 10 tests. Indicates the standard deviation of the number of wrong questions in the last 10 tests. Indicates the number of page rollbacks in a single learning session, d indicates the time spent on the current page, Indicates the average residence time of similar resources. Indicates the average number of page rollbacks during a single learning session for other learners on the platform; The calculation formula for the resource type efficiency characteristics is: ,in, Indicates the resource type performance characteristics, represents the learning time of the learner on resource type r, Indicates the learning time of learners on all resource types, represents the average of the learners’ test scores, Indicates the full score of the test; A task capability map construction module is used to construct a task capability map based on the association between job tasks and capability indicators, capability indicators and course modules, and course modules and knowledge points predefined on the learning platform. The task capability map includes job task nodes, capability indicator nodes, course module nodes, and knowledge point nodes, as well as the associated edges between the nodes; Calculate the mastery score of each capability indicator node based on the test results; Extract sub-graphs from the task ability graph based on mastery scores and behavioral inertia parameters; The recommended path generation module is used to generate the optimal learning recommendation path based on the behavioral inertia parameters and the subgraph map according to the path recommendation engine. The path recommendation engine is used to generate Q candidate learning recommendation paths and select the optimal learning recommendation path as the recommendation result based on the path fitness score. The candidate learning recommendation path is a sequence of course module nodes and knowledge point nodes arranged in sequence, and Q is a positive integer.

2. The learning path intelligent recommendation system based on user behavior big data analysis according to claim 1 is characterized in that: The relationships in the task capability map include: There is a one-to-many relationship between job task nodes and capability indicator nodes; There is a one-to-many association between capability indicator nodes and course module nodes; There is a one-to-many association between course module nodes and knowledge point nodes; Directed edges are established between knowledge point nodes based on the previous and next dependencies of the knowledge system; Directed edges are generated between nodes of different types to construct a nested graph structure of job tasks-competency indicators-course modules-knowledge points.

3. The learning path intelligent recommendation system based on user behavior big data analysis according to claim 1 is characterized in that: The mastery score of the capability indicator node is calculated based on the test scores associated with the capability indicator node, wherein the test scores of the capability indicator node are normalized and weighted summed to obtain the mastery score.

4. The learning path intelligent recommendation system based on user behavior big data analysis according to claim 3 is characterized in that: Based on the mastery score and behavioral inertia parameters, a subgraph is extracted from the task capability graph. The specific steps include: S201, screening ability indicator nodes whose mastery score is greater than a first preset threshold and whose learning rhythm stability coefficient is greater than a second preset threshold; S202, extracting, based on the screened capability indicator nodes, the job task nodes and course module nodes connected thereto, as well as the knowledge point nodes connected to the course module nodes; S203, obtaining a subgraph map based on the screened capability indicator nodes and the extracted job task nodes, course module nodes and knowledge point nodes, as well as the edges between the nodes.

5. The learning path intelligent recommendation system based on user behavior big data analysis according to claim 4 is characterized in that: The recommended path generation module imposes the following internal constraints during the candidate learning recommended path generation process executed by the path recommendation engine: The order of course module nodes and knowledge point nodes is the learning order between course module nodes and knowledge point nodes in the subgraph; There are edges between adjacent knowledge point nodes in the learner's candidate learning recommendation path; The estimated total learning time of the learner's candidate recommended learning paths does not exceed a fourth preset threshold.

6. The learning path intelligent recommendation system based on user behavior big data analysis according to claim 1 is characterized in that: The recommended path generation module further includes: a module for receiving intervention information input from an external port as an external constraint imposed by the path recommendation engine during the generation of candidate learning recommended paths, wherein the intervention information includes: Specify the mandatory priority learning order of course module nodes; Specify that the candidate learning recommendation path contains specific knowledge point nodes; The estimated total learning time of the designated candidate learning recommendation path is not less than a fifth preset threshold.

7. The learning path intelligent recommendation system based on user behavior big data analysis according to claim 1 is characterized in that: The recommended path generation module performs structured coding on the candidate learning recommended paths, and the specific steps include: S301: hierarchically encode the course module nodes and knowledge point nodes in the candidate learning recommendation path according to their dependency relationships in the subgraph to generate a node sequence with a topological sequence number; S302 , mapping the resource type efficiency feature in the behavioral inertia parameter into a node weight label, and forming a coding vector with the node sequence generated in S301 .

8. The learning path intelligent recommendation system based on user behavior big data analysis according to claim 1 is characterized in that: The calculation process of the path fitness score includes: S401: Detect nodes in the node sequence that violate the dependency relationship in the subgraph graph as conflict nodes, assign penalty values ​​to the conflict nodes, and generate a sequence rationality score based on the hierarchical depth of the conflict nodes; S402, setting the difficulty level of course module nodes and knowledge point nodes, counting the node pairs in the candidate learning recommendation path whose difficulty level difference exceeds 2 levels, and obtaining the difficulty gradient smoothness score through exponential decay calculation; S403 , performing weighted addition on the sequence rationality score and the difficulty gradient smoothness score to obtain a path fitness score.

9. The learning path intelligent recommendation system based on user behavior big data analysis according to claim 1 is characterized in that: The step of selecting the optimal learning recommendation path based on the path fitness score by the recommended path generation module includes: S501, sorting the Q candidate learning recommendation paths in descending order according to the path fitness scores, and retaining a preset number of candidate learning recommendation paths; S502: Perform a first mutation operation on the retained candidate learning recommendation path with a first preset probability. The first mutation operation includes: Randomly select a course module node in the candidate learning recommendation path; Retrieve knowledge point nodes associated with the course module nodes but not in the candidate learning recommendation path from the subgraph map, and filter out knowledge point nodes whose resource type effectiveness characteristics are higher than a sixth preset threshold; Replace the knowledge point nodes connected to the course module nodes in the node sequence with the filtered knowledge point nodes; S503, recalculating the path fitness score for the candidate learning recommendation path after the first mutation operation; S504, repeatedly execute S501 to S503 until the improvement of the highest path fitness score among the candidate learning paths in consecutive U rounds of iterations does not exceed the seventh preset threshold, and the candidate learning recommendation path with the highest path fitness score in the U round is used as the optimal learning recommendation path, where U is a positive integer.

Images