A smart course adaptive learning system for digital education

Abstract

The present invention discloses a digital education smart course adaptive learning system, including a data acquisition module, a data preprocessing module, a core learner group determination module, a cluster processing module and an adaptive learning recommendation module. The present invention belongs to the field of learning recommendation technology, and specifically refers to a digital education smart course adaptive learning system. This scheme adopts a unique dynamic weight update mechanism, and dynamically adjusts the K value to flexibly respond to the differences in learner distribution; resists interference from abnormal learners through the balance of local and global neighboring strengths; ensures the accuracy of learning resource recommendations; introduces the weight adjustment mechanism into learner distance calculation; the feedback mechanism adjusts the learner's difference distance in real time by introducing the learner's feedback score; updates learner characteristics based on self-supervised learning, dynamically updates clustering results, and better adapts to learner behavior changes; thereby improving the efficiency of adaptive learning and improving learning effects.

Description

Technical Field

[0001] The present invention relates to the technical field of learning recommendation, and in particular to a smart course adaptive learning system for digital education. Background Art

[0002] The smart course adaptive learning system is a personalized learning platform based on data-driven and artificial intelligence technology. It aims to dynamically adjust learning resources, learning paths, and learning content recommendations based on learners' personalized learning behaviors, needs, and feedback, ultimately improving learners' learning efficiency and learning effects. However, the general smart course adaptive learning system has static learner classification, which makes it impossible to adjust learners' similarity weights in time and cannot handle learner groups with different proximity strengths; the general smart course adaptive learning system has learner feedback mechanisms that have little impact on system recommendations and cannot handle dynamic learner characteristics, which leads to low learner learning effects and cannot meet learners' actual needs. Summary of the invention

[0003] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a smart course adaptive learning system for digital education. In view of the problem that the general smart course adaptive learning system has static learner classification, which leads to the inability to timely adjust the learner similarity weight and the inability to handle learner groups with different proximity strengths, the present solution adopts a unique dynamic weight update mechanism and dynamically adjusts the K value to flexibly cope with the differences in learner distribution; resists the interference of abnormal learners by balancing local and global proximity strengths; ensures the accuracy of learning resource recommendations; in view of the problem that the learner feedback mechanism of the general smart course adaptive learning system has little effect on the system recommendation and cannot handle dynamic learner characteristics, which leads to low learning effect of learners and cannot meet the actual needs of learners, the present solution introduces a weight adjustment mechanism into the learner distance calculation; the feedback mechanism adjusts the learner difference distance in real time by introducing the learner's feedback score; improves personalized recommendations to ensure that the recommended content meets the actual needs of learners; updates learner characteristics based on self-supervised learning, dynamically updates clustering results, and better adapts to learner behavior changes; thereby improving the efficiency of adaptive learning and improving learning effects.

[0004] The technical solution adopted by the present invention is as follows: A digital education smart course adaptive learning system provided by the present invention includes a data acquisition module, a data preprocessing module, a core learner group determination module, a clustering processing module and an adaptive learning recommendation module;

[0005] The data collection module collects the learning behavior data of the learners;

[0006] The data preprocessing module performs data cleaning, data conversion and standardization on the collected data;

[0007] The core learner group determination module adopts a dynamic weight update mechanism and dynamically adjusts the K value; the core learner group is determined by balancing the local and global proximity strengths;

[0008] The clustering processing module introduces a weight adjustment mechanism into learner distance calculation; the feedback mechanism adjusts learner difference distances in real time by introducing learner feedback scores; learner features are updated based on self-supervised learning, and clustering results are dynamically updated;

[0009] The adaptive learning recommendation module makes adaptive school recommendations for learners based on clustering results.

[0010] Furthermore, in the data collection module, the learner's learning behavior data includes learning behavior data, course progress data, test data, educational background data, learning resource data and the learner's feedback score on the current learning recommendation; the learning behavior data includes learning time, learning frequency and learning resource preference; the course progress data includes course module completion, learning module completion time and the number of repeated learning times of course chapters; the test data includes test scores, error rate and test time; the learning resource data includes video courses, text materials, interactive exercises and audio materials used by learners in the learning process; the learning resource data is used as a data label, and the learner's feedback score on the current learning recommendation is only used to calculate the difference distance, and neither participates in data calculation.

[0011] Furthermore, in the data preprocessing module, the data cleaning is to process duplicate values ​​and outliers; the data conversion is to convert the data into a vector form; and the standardization processing is to standardize the data based on the maximum and minimum normalization method.

[0012] Furthermore, the core learner group determination module includes the following contents:

[0013] Calculate the local proximity strength of learners; Calculate the local proximity strength ρ of each learner based on the learner behavior data i , the local neighborhood strength reflects the behavioral similarity between each learner and its neighboring learners, expressed as: ; and dynamically update the weights, expressed as: ; where j is the neighbor index; KNN i is the set of neighbors of the i-th learner; is the weight between the i-th learner and the j-th learner; is the updated weight; d ij is the Euclidean distance between learners; α is the weight change factor; F i and F jare the feature vectors of the i-th learner and the j-th learner respectively; is the amplitude variable; is the local neighbor strength of the jth learner; maxρ is the maximum local neighbor strength; maxd is the maximum distance;

[0014] Determine the number of K nearest neighbors; the initial K value is determined according to the size of the data set and is expressed as: ; Dynamically adjust the K value at each iteration, expressed as: ; Where n is the number of learners; is the updated K value; is the standard deviation of the distance between the learner and its neighbors; is the maximum distance standard deviation in the entire data set; γ is a dynamic adjustment parameter;

[0015] Determine the core learner group; select the learner with the highest local proximity strength as the initial core learner, expressed as: ; Sort the learners from high to low according to the proximity strength, and calculate the K nearest neighbors of each learner, expressed as: ; To judge whether a learner can become a core learner, it is expressed as: ; Among them, L it is the core learner selected in the tth iteration; α1 is the difference weight; block i is sorting; L it+1 is the core learner selected in the t+1th iteration; S i is to judge whether to join; τ is the allowed overlap ratio; is the reference value of the global neighborhood strength; L i is the i-th learner.

[0016] Furthermore, the cluster processing module specifically includes the following contents:

[0017] Calculate the direct distance: Calculate the direct distance between the learner and the core learner using the following formula:

[0018] ;

[0019] In the formula, is the Euclidean distance between the ith learner and the Lth core learner; is the weight between the i-th learner and the L-th core learner; is the direct distance between the ith learner and the Lth core learner;

[0020] Calculate the difference distance; for each learner, based on the connectivity distance between local centers, calculate the difference distance between the learner and the core learner group; the formula used is as follows:

[0021] ;

[0022] In the formula, is the difference distance between the i-th learner and the L-th core learner; β is the influence coefficient; f iL is the learner's feedback score on the current learning recommendation, initially 0;

[0023] Clustering; assign non-core learners to the core learner group based on the minimum difference distance; dynamically update the core learners through self-supervised learning; in each iteration, use the self-supervised learning model to compare the learner's characteristics, update the core learner selection criteria, and use the contrast loss function, which is expressed as: ; Where Lc is the contrast loss function; is the similarity label between learners, which is 1 if they belong to the same cluster, otherwise it is 0; is the distance between learners; m is the control threshold; specifically: after each iteration, the learner's feature vector is updated by gradient descent based on the contrast loss function, and the local proximity strength is recalculated based on the updated learner's feature vector; a judgment threshold is set in advance, and the distance between the core learners is calculated. When the distance is lower than the judgment threshold, a merge operation is performed; the core learners are selected and assigned again until the clustering converges and the clustering is completed.

[0024] Furthermore, the adaptive learning recommendation module recommends the learning resource data of the core learners to the non-core learners in the same cluster based on the clustering result.

[0025] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0026] (1) In view of the problem that the general intelligent course adaptive learning system has static learner classification, which makes it impossible to adjust the learner similarity weight in time and cannot handle learner groups with different proximity strengths, this scheme adopts a unique dynamic weight update mechanism and dynamically adjusts the K value to flexibly respond to the differences in learner distribution; resists the interference of abnormal learners by balancing local and global proximity strengths; and ensures the accuracy of learning resource recommendations.

[0027] (2) In view of the problem that the learner feedback mechanism of general intelligent course adaptive learning systems has little impact on system recommendations and cannot handle dynamic learner characteristics, which leads to low learning effects and cannot meet learners' actual needs, this scheme introduces a weight adjustment mechanism into learner distance calculation; the feedback mechanism adjusts learners' difference distances in real time by introducing learners' feedback scores; improves personalized recommendations to ensure that the recommended content meets learners' actual needs; updates learner characteristics based on self-supervised learning, dynamically updates clustering results, and better adapts to learners' behavioral changes; thereby improving the efficiency of adaptive learning and improving learning effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 A flowchart of a smart course adaptive learning system for digital education provided by the present invention;

[0029] Figure 2 Schematic diagram of the process of identifying modules for core learner groups;

[0030] Figure 3 Schematic diagram of the clustering processing module.

[0031] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0032] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0033] In the description of the present invention, it should be understood that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as limiting the present invention.

[0034] Example 1, see Figure 1 , a digital education smart course adaptive learning system provided by the present invention includes a data acquisition module, a data preprocessing module, a core learner group determination module, a clustering processing module and an adaptive learning recommendation module;

[0035] The data acquisition module collects the learning behavior data of the learners; and sends the data to the data preprocessing module;

[0036] The data preprocessing module receives the data sent by the data acquisition module; performs data cleaning, data conversion and standardization on the collected data; and sends the data to the core learner group determination module;

[0037] The core learner group determination module receives data sent by the data preprocessing module; adopts a dynamic weight update mechanism and dynamically adjusts the K value; determines the core learner group by balancing the local and global proximity strengths; and sends the data to the clustering processing module;

[0038] The cluster processing module receives data sent by the core learner group determination module; introduces a weight adjustment mechanism into learner distance calculation; the feedback mechanism adjusts learner difference distances in real time by introducing learner feedback scores; updates learner features based on self-supervised learning and dynamically updates clustering results; and sends data to the adaptive learning recommendation module;

[0039] The adaptive learning recommendation module receives the data sent by the clustering processing module; and makes adaptive school recommendations to the learners based on the clustering results.

[0040] Example 2, see Figure 1 This embodiment is based on the above embodiment. In the data collection module, the learner's learning behavior data includes learning behavior data, course progress data, test data, educational background data, learning resource data and the learner's feedback score on the current learning recommendation; the learning behavior data includes learning time, learning frequency and learning resource preference; the course progress data includes course module completion, learning module completion time and the number of repeated learning times of course chapters; the test data includes test scores, error rate and test time; the learning resource data includes video courses, text materials, interactive exercises and audio materials used by learners in the learning process; the learning resource data is used as a data label, and the learner's feedback score on the current learning recommendation is only used to calculate the difference distance, and neither participates in data calculation.

[0041] Example 3, see Figure 1 This embodiment is based on the above embodiment. In the data preprocessing module, data cleaning is to process duplicate values ​​and outliers; data conversion is to convert data into vector form; and standardization is to standardize data based on the maximum and minimum normalization method.

[0042] Example 4, see Figure 1 and Figure 2 This embodiment is based on the above embodiment, and the core learner group determination module includes the following contents:

[0043] Calculate the local proximity strength of learners; Calculate the local proximity strength ρ of each learner based on the learner behavior data i , the local neighborhood strength reflects the behavioral similarity between each learner and its neighboring learners, expressed as: ; and dynamically update the weights, expressed as: ; where j is the neighbor index; KNN i is the set of neighbors of the i-th learner; is the weight between the i-th learner and the j-th learner; is the updated weight; d ij is the Euclidean distance between learners; α is the weight change factor; F i and F j are the feature vectors of the i-th learner and the j-th learner respectively; is the amplitude variable; is the local neighbor strength of the jth learner; maxρ is the maximum local neighbor strength; maxd is the maximum distance;

[0044] Determine the number of K nearest neighbors; the initial K value is determined according to the size of the data set and is expressed as: ; Dynamically adjust the K value at each iteration, expressed as: ; Where n is the number of learners; is the updated K value; is the standard deviation of the distance between the learner and its neighbors; is the maximum distance standard deviation in the entire data set; γ is a dynamic adjustment parameter;

[0045] Determine the core learner group; select the learner with the highest local proximity strength as the initial core learner, expressed as: ; Sort the learners from high to low according to the proximity strength, and calculate the K nearest neighbors of each learner, expressed as: ; To judge whether a learner can become a core learner, it is expressed as: ; Among them, L it is the core learner selected in the tth iteration; α1 is the difference weight; block i is sorting; L it+1 is the core learner selected in the t+1th iteration; S i is to judge whether to join; τ is the allowed overlap ratio; is the reference value of the global neighborhood strength; L i is the i-th learner.

[0046] By performing the above operations, in order to solve the problem that the general intelligent course adaptive learning system has static learner classification, which makes it impossible to adjust the learner similarity weight in time and cannot handle learner groups with different proximity strengths, this solution adopts a unique dynamic weight update mechanism and dynamically adjusts the K value to flexibly respond to the differences in learner distribution; resists the interference of abnormal learners by balancing local and global proximity strengths; and ensures the accuracy of learning resource recommendations.

[0047] Example 5, see Figure 1 and Figure 3 This embodiment is based on the above embodiment, and the cluster processing module specifically includes the following contents:

[0048] Calculate the direct distance: Calculate the direct distance between the learner and the core learner using the following formula:

[0049] ;

[0050] In the formula, is the Euclidean distance between the ith learner and the Lth core learner; is the weight between the i-th learner and the L-th core learner; is the direct distance between the ith learner and the Lth core learner;

[0051] Calculate the difference distance; for each learner, based on the connectivity distance between local centers, calculate the difference distance between the learner and the core learner group; the formula used is as follows:

[0052] ;

[0053] In the formula, is the difference distance between the i-th learner and the L-th core learner; β is the influence coefficient; f iL is the learner's feedback score on the current learning recommendation, initially 0;

[0054] Clustering; assign non-core learners to the core learner group based on the minimum difference distance; dynamically update the core learners through self-supervised learning; in each iteration, use the self-supervised learning model to compare the learner's characteristics, update the core learner selection criteria, and use the contrast loss function, which is expressed as: ; Where Lc is the contrast loss function; is the similarity label between learners, which is 1 if they belong to the same cluster, otherwise it is 0; is the distance between learners; m is the control threshold; specifically: after each iteration, the learner's feature vector is updated by gradient descent based on the contrast loss function, and the local proximity strength is recalculated based on the updated learner's feature vector; a judgment threshold is set in advance, and the distance between the core learners is calculated. When the distance is lower than the judgment threshold, a merge operation is performed; the core learners are selected and assigned again until the clustering converges and the clustering is completed.

[0055] By performing the above operations, in view of the problem that the learner feedback mechanism of the general smart course adaptive learning system has little effect on the system recommendation and cannot handle dynamic learner characteristics, which leads to low learning effect of learners and cannot meet the actual needs of learners, this solution introduces a weight adjustment mechanism into the learner distance calculation; the feedback mechanism adjusts the learner difference distance in real time by introducing the learner's feedback score; improves personalized recommendations to ensure that the recommended content meets the actual needs of learners; updates learner characteristics based on self-supervised learning, dynamically updates clustering results, and better adapts to learners' behavioral changes; thereby improving the efficiency of adaptive learning and improving learning effects.

[0056] Example 6, see Figure 1 ,This embodiment is based on the above embodiment, and the adaptive learning recommendation module recommends the ,learning resource data of the core learners to the non-core learners in the same cluster based on ,clustering results.

[0057] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0058] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0059] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. A digital education intelligent course adaptive learning system, characterized by: The system includes a data collection module, a data preprocessing module, a core learner group determination module, a clustering processing module and an adaptive learning recommendation module; The data collection module collects the learning behavior data of the learners; The data preprocessing module performs data cleaning, data conversion and standardization on the collected data; The core learner group determination module adopts a dynamic weight update mechanism and dynamically adjusts the K value; the core learner group is determined by balancing the local and global proximity strengths; The clustering processing module introduces a weight adjustment mechanism into learner distance calculation; the feedback mechanism adjusts learner difference distances in real time by introducing learner feedback scores; learner features are updated based on self-supervised learning, and clustering results are dynamically updated; The adaptive learning recommendation module makes adaptive school recommendations for learners based on clustering results; The core learner group determination module includes calculating the local proximity strength of the learners; The learner behavior data calculates the local neighborhood strength ρ of each learner i , the local neighborhood strength reflects the behavioral similarity between each learner and its neighboring learners, expressed as: ; and dynamically update the weights, expressed as: ; where j is the neighbor index; KNN i is the set of neighbors of the i-th learner; is the weight between the i-th learner and the j-th learner; is the updated weight; d ij is the Euclidean distance between learners; α is the weight change factor; F i and F j are the feature vectors of the i-th learner and the j-th learner respectively; is the amplitude variable; is the local neighbor strength of the jth learner; maxρ is the maximum local neighbor strength; maxd is the maximum distance; The cluster processing module specifically includes the following contents: Calculate the direct distance: Calculate the direct distance between the learner and the core learner using the following formula: ； In the formula, is the Euclidean distance between the ith learner and the Lth core learner; is the weight between the i-th learner and the L-th core learner; is the direct distance between the ith learner and the Lth core learner; Calculate the difference distance; for each learner, calculate the difference distance between the learner and the core learner group based on the connectivity distance between the local centers; the formula used is as follows: ； In the formula, is the difference distance between the i-th learner and the L-th core learner; β is the influence coefficient; f iL is the learner's feedback score on the current learning recommendation, initially 0; Clustering; assign non-core learners to the core learner group based on the minimum difference distance; dynamically update the core learners through self-supervised learning; in each iteration, use the self-supervised learning model to compare the learner's characteristics, update the core learner selection criteria, and use the contrast loss function, which is expressed as: ; Where Lc is the contrast loss function; is the similarity label between learners, which is 1 if they belong to the same cluster, otherwise it is 0; is the distance between learners; m is the control threshold; specifically: after each iteration, the learner's feature vector is updated by gradient descent based on the contrast loss function, and the local proximity strength is recalculated based on the updated learner's feature vector; a judgment threshold is set in advance, and the distance between each core learner is calculated. When the distance is lower than the judgment threshold, a merge operation is performed; core learners are selected and assigned again until clustering converges and clustering is completed; The core learner group identification module also includes the following: Determine the number of K nearest neighbors; the initial K value is determined according to the size of the data set, expressed as: ; Dynamically adjust the K value at each iteration, expressed as: ; Where n is the number of learners; is the updated K value; is the standard deviation of the distance between the learner and its neighbors; is the maximum distance standard deviation in the entire data set; γ is a dynamic adjustment parameter; Determine the core learner group; select the learner with the highest local proximity strength as the initial core learner, expressed as: ; Sort the learners from high to low according to the proximity strength, and calculate the K nearest neighbors of each learner, expressed as: ; To judge whether a learner can become a core learner, it is expressed as: ; Among them, L it is the core learner selected in the tth iteration; α1 is the difference weight; block i is sorting; L it+1 is the core learner selected in the t+1th iteration; S i is to judge whether to join; τ is the allowed overlap ratio; is the reference value of global proximity strength; L i is the i-th learner.

2. According to claim 1, a digital education intelligent course adaptive learning system is characterized by: In the data collection module, the learner's learning behavior data includes learning behavior data, course progress data, test data, educational background data, learning resource data and learner's feedback score on current learning recommendations; the learning behavior data includes learning time, learning frequency and learning resource preference; the course progress data includes course module completion, learning module completion time and the number of repeated learning of course chapters; the test data includes test scores, error rate and test time; the learning resource data includes video courses, text materials, interactive exercises and audio materials used by learners in the learning process; The learning resource data is used as the data label, and the learner's feedback score on the current learning recommendation is only used to calculate the difference distance and does not participate in data calculation.

3. The digital education intelligent course adaptive learning system according to claim 1, characterized in that: The adaptive learning recommendation module recommends the learning resource data of core learners to non-core learners in the same cluster based on the clustering results.

4. The digital education intelligent course adaptive learning system according to claim 1, characterized in that: In the data preprocessing module, the data cleaning is to process duplicate values ​​and outliers; the data conversion is to convert the data into vector form; and the standardization processing is to standardize the data based on the maximum and minimum normalization method.

Images