Intelligent teaching platform resource recommendation management method

Abstract

The invention belongs to the technical field of teaching platform resource recommendation, and discloses an intelligent teaching platform resource recommendation management method. Comprising the steps of collecting explicit data, implicit data and situation data; according to the implicit data, obtaining a long-term interest value and a short-term interest value of the user for the online courses, according to the long-term interest value and the short-term interest value, obtaining a comprehensive interest value of the user for the online courses, and according to the comprehensive interest value, recommending the corresponding online courses to the user; obtaining a concentration value of the user for the offline course and a knowledge level value of the user according to the situation data, setting a concentration value threshold range, obtaining a knowledge level grade of the user according to the knowledge level value of the user, and when the concentration value exceeds the concentration value threshold range, recommending the corresponding offline course to the user according to the explicit data. When the concentration value is within the concentration value threshold range, recommending a corresponding offline course to the user according to the knowledge level grade; and the accuracy and adaptability of teaching platform resource recommendation are further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of teaching platform resource recommendation, and more specifically, to a resource recommendation and management method for an intelligent teaching platform. Background Art

[0002] Patent application publication number CN117726082A discloses a teaching resource recommendation method, device, electronic device, and readable storage medium. The method includes: collecting user feedback information from target users when using a teaching resource recommendation system, receiving the target learning progress for each learning task in the target subject set for the target user and the target knowledge point the target user is currently learning; determining the target user's current learning progress and the target user's preference for each type of teaching resource based on the user feedback information; evaluating the utility of each teaching resource for each learning task in the target subject and the consumption incurred by the target user when using the teaching resource; calculating the teaching increment of the teaching resource for the target knowledge point based on the utility, consumption, target learning progress, current learning progress, preference, and the strength of the association between the target knowledge point and the teaching resource; and recommending the target teaching resource to the target user based on the teaching increment. This method improves the accuracy of teaching resource recommendations.

[0003] However, in the process of recommending resources on the teaching platform, the user base is large, and each user's long-term interests, short-term interests and knowledge level are different. The existing technology combines long-term interests and short-term interests according to the weighted average summation method, and the weights of long-term interests and short-term interests remain unchanged. If the user has not watched online courses for a long time, the weight of long-term interests should increase. The existing technology does not change the weight, which will lead to inaccurate recommendations of teaching platform resources. If the user has frequently watched online courses recently, the weight of short-term interests should increase. The existing technology does not change the weight, which will also lead to inaccurate recommendations of teaching platform resources. If the teaching platform resources are recommended without considering the user's knowledge level, the difficulty of the recommended offline courses may be too large or too small relative to the user, thereby reducing the accuracy and adaptability of the teaching platform resource recommendation.

[0004] In view of this, the present invention proposes a resource recommendation and management method for an intelligent teaching platform to solve the above problems. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solution: a resource recommendation and management method for an intelligent teaching platform, comprising:

[0006] Step S1: Collect explicit data, implicit data and contextual data;

[0007] Step S2: Obtain the user's long-term interest value and short-term interest value for the online course based on implicit data, obtain the user's comprehensive interest value for the online course based on the long-term interest value and the short-term interest value, and recommend corresponding online courses to the user based on the comprehensive interest value; wherein, for the comprehensive interest value, the user's boredom coefficient is obtained by using the Sigmoid function based on the time when the user watched the online course most recently. Based on the user's boredom coefficient, the long-term interest value and the short-term interest value are fused through adaptive adjustment to obtain the user's comprehensive interest value for the online course;

[0008] Step S3: Obtain the user's concentration value for the offline course and the user's knowledge level value based on the contextual data, set a concentration value threshold range, and obtain the user's knowledge level level based on the user's knowledge level value;

[0009] Step S4: When the concentration value exceeds the concentration value threshold range, the corresponding offline courses are recommended to the user based on the explicit data; when the concentration value is within the concentration value threshold range, the corresponding offline courses are recommended to the user based on the knowledge level.

[0010] Furthermore, the explicit data includes user interest tags;

[0011] The implicit data includes the viewing time of online courses, comment data on online courses, and the accuracy rate of exercises corresponding to online courses when users study online. The contextual data includes video recording data of users studying offline and the scores of exercises corresponding to offline courses.

[0012] Furthermore, the method for obtaining the user's long-term interest value and short-term interest value for the online course based on implicit data includes:

[0013] Extract the online course viewing time, online course comment data, and online course corresponding exercise accuracy of users during online learning within T days from the current time, use a two-way matching algorithm to segment the comment data within T days from the current time, collect all the segmented words to obtain a long-term two-way comment word set, set a diverse word database, and store p_v words in the diverse word database. The diverse word database is provided with a positive word storage space, a medium word storage space, and a negative word storage space, which respectively store words with positive evaluation, words with medium evaluation, and words with negative evaluation;

[0014] Compare the words in the long-term two-way review word set with the words stored in the diverse word database, and obtain the long-term review impact factor based on the comparison result;

[0015] An improved weighted average summation method is used to obtain the long-term interest value of the user for the online course based on the viewing duration of the online course during the user's online learning within T days from the current time, the long-term comment influence factor, and the correct rate of answering questions for the corresponding exercises of the online course;

[0016] The short-term interest value of the user for the online course is obtained based on the viewing duration of the online course during the user's online learning within N days from the current time, the comment data on the online course, and the correct rate of answering questions for the corresponding exercises of the online course.

[0017] Furthermore, the method for obtaining the long-term comment influence factor according to the comparison result includes:

[0018] If a word in the long-term two-way comment word set is the same as any positive evaluation word stored in the positive word storage space, this word is recorded as a positive word; if a word in the long-term two-way comment word set is the same as any medium evaluation word stored in the medium word storage space, this word is recorded as a medium word; if a word in the long-term two-way comment word set is the same as any negative evaluation word stored in the negative word storage space, this word is recorded as a negative word;

[0019] Obtain the number of positive words, the number of medium words, and the number of negative words in the long-term two-way comment word set, and fuse the number of positive words, the number of medium words, and the number of negative words in the long-term two-way comment word set through adaptive adjustment to obtain the long-term comment influence factor.

[0020] Furthermore, the method for obtaining the short-term interest value of the user for the online course includes:

[0021] Use a two-way matching algorithm to segment the comment data within N days from the current time, where N < T, and collect all the words obtained by segmentation to obtain a short-term two-way comment word set;

[0022] Compare the words in the short-term two-way comment word set with the words stored in the diverse word database, and obtain the short-term comment influence factor according to the comparison result;

[0023] An improved weighted average summation method is used to obtain the short-term interest value of the user for the online course based on the viewing duration of the online course during the user's online learning within N days from the current time, the short-term comment influence factor, and the correct rate of answering questions for the corresponding exercises of the online course.

[0024] Furthermore, the method for recommending the corresponding online course to the user according to the comprehensive interest value includes:

[0025] Set a comprehensive interest value threshold, compare the user's comprehensive interest value for online courses with the comprehensive interest value threshold, obtain online courses corresponding to comprehensive interest values ​​greater than or equal to the comprehensive interest value threshold, and record them as interested online courses;

[0026] Obtain all knowledge points in the online courses and connect the corresponding knowledge points based on the association relationships between the knowledge points, including the relationship between the prior knowledge, the hierarchical relationship, and the similar concept relationship;

[0027] If the association relationship between two knowledge points is a pre-knowledge relationship, the two knowledge points are connected and the pre-knowledge relationship is marked on the connection line. If the association relationship between two knowledge points is a hierarchical relationship, the two knowledge points are connected and the hierarchical relationship is marked on the connection line. If the association relationship between two knowledge points is a similar concept relationship, the two knowledge points are connected and the similar concept relationship is marked on the connection line.

[0028] Obtain the knowledge points connected to the knowledge points in the online courses of interest, record them as knowledge points of interest, and recommend online courses corresponding to the knowledge points of interest to users.

[0029] Furthermore, the method for obtaining the user's concentration value for offline courses and the user's knowledge level value based on the contextual data includes:

[0030] Obtain the total number of blinks of the user from the video recording data of the user's offline learning, and use the total number of blinks of the user as the user's concentration value for the offline course;

[0031] The user's knowledge level value is obtained by summing up the scores of the corresponding exercises in the offline courses.

[0032] Furthermore, the method for setting the concentration value threshold range includes:

[0033] Get the normal blink frequency range per minute and offline course duration;

[0034] Calculate the product of the minimum value in the normal number of blinks in one minute and the offline course duration to obtain the minimum value in the concentration value threshold range. Calculate the product of the maximum value in the normal number of blinks in one minute and the offline course duration to obtain the maximum value in the concentration value threshold range.

[0035] Furthermore, the method for obtaining the user's knowledge level grade according to the user's knowledge level value includes:

[0036] The knowledge level levels include first level knowledge level, second level knowledge level and third level knowledge level;

[0037] Set the knowledge level value threshold range;

[0038] When the user's knowledge level value is less than the minimum value in the knowledge level value threshold range, the user's knowledge level relative to the offline course is the first level knowledge level;

[0039] When the user's knowledge level value is less than or equal to the maximum value in the knowledge level value threshold range and greater than or equal to the minimum value in the knowledge level value threshold range, the user's knowledge level level relative to the offline course is the second level knowledge level;

[0040] When the user's knowledge level value is greater than the maximum value in the knowledge level value threshold range, the user's knowledge level rank relative to the offline course is the third knowledge level.

[0041] Furthermore, when the concentration value exceeds the concentration value threshold range, the method of recommending corresponding offline courses to the user based on explicit data, and when the concentration value is within the concentration value threshold range, the method of recommending corresponding offline courses to the user based on the knowledge level includes:

[0042] When the concentration value exceeds the concentration value threshold range, the offline courses corresponding to the user's interest tags are recommended to the user;

[0043] When the concentration value is within the concentration value threshold range, the corresponding offline courses are recommended to the user based on the knowledge level;

[0044] Offline courses for each subject are divided into difficulty levels, including level 1, level 2, and level 3.

[0045] When the user's knowledge level is at level one relative to the offline course, the subject corresponding to the offline course is obtained and recorded as a level one subject. The offline course of level one difficulty corresponding to the level one subject is recommended to the user.

[0046] When the user's knowledge level is at level 2 relative to the offline course, the subject corresponding to the offline course is obtained and recorded as a level 2 subject, and offline courses of level 2 difficulty corresponding to the level 2 subject are recommended to the user;

[0047] When the user's knowledge level relative to the offline course is at level three, the subject corresponding to the offline course is obtained and recorded as a level three subject, and an offline course of level three difficulty corresponding to the level three subject is recommended to the user.

[0048] The technical effects and advantages of the intelligent teaching platform resource recommendation management method of the present invention are as follows:

[0049] 1. Obtaining users' long-term and short-term interest in online courses based on implicit data. When obtaining long-term interest, users' interests change slowly over time. Early learning behaviors should be given lower weights, while recent learning behaviors should be given higher weights. Therefore, a time-decay weighting coefficient is introduced to make the obtained long-term interest values ​​more accurate, thereby improving the accuracy of resource recommendations on the teaching platform.

[0050] 2. A user's comprehensive interest is related to long-term and short-term interests. The user's boredom coefficient is obtained based on the time when the user last watched an online course. The user's comprehensive interest value is adjusted based on the user's boredom coefficient. If the user watches online courses every day, the smaller the user's boredom coefficient, the greater the impact of the short-term interest value on the comprehensive interest value. If the user has not watched online courses for one month, the larger the user's boredom coefficient, the greater the impact of the long-term interest value on the comprehensive interest value. By adjusting the user's comprehensive interest value based on the user's boredom coefficient, the problem of inaccurate comprehensive interest value is solved, making the obtained comprehensive interest value more real-time and accurate, thereby improving the accuracy of resource recommendations on the teaching platform;

[0051] 3. Obtain the user's concentration value for offline courses and the user's knowledge level value based on contextual data, set the concentration value threshold range, and obtain the user's knowledge level grade based on the user's knowledge level value. When the concentration value exceeds the concentration value threshold range, recommend the corresponding offline course to the user based on the explicit data; when the concentration value is within the concentration value threshold range, recommend the corresponding offline course to the user based on the knowledge level grade; recommend offline courses of corresponding difficulty to the user based on the user's knowledge level grade, thereby increasing the user's interest in the offline course, increasing the user's motivation to learn, and thereby improving the adaptability of resource recommendations on the teaching platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 This is a schematic diagram of a resource recommendation management method for an intelligent teaching platform according to the present invention;

[0053] Figure 2 This is a schematic diagram of a resource recommendation management system for an intelligent teaching platform according to the present invention;

[0054] Figure 3 Schematic diagram of the method for obtaining a user's short-term interest value in an online course according to the present invention. DETAILED DESCRIPTION

[0055] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0056] Example 1

[0057] See also Figure 1 and Figure 3 As shown, the resource recommendation and management method of an intelligent teaching platform described in this embodiment includes:

[0058] Step S1: Collect explicit data, implicit data and contextual data;

[0059] Step S2: Obtain the user's long-term interest value and short-term interest value for the online course based on implicit data, obtain the user's comprehensive interest value for the online course based on the long-term interest value and the short-term interest value, and recommend corresponding online courses to the user based on the comprehensive interest value; wherein, for the comprehensive interest value, the user's boredom coefficient is obtained by using the Sigmoid function based on the time when the user watched the online course most recently. Based on the user's boredom coefficient, the long-term interest value and the short-term interest value are fused through adaptive adjustment to obtain the user's comprehensive interest value for the online course;

[0060] Step S3: Obtain the user's concentration value for the offline course and the user's knowledge level value based on the contextual data, set a concentration value threshold range, and obtain the user's knowledge level level based on the user's knowledge level value;

[0061] Step S4: When the concentration value exceeds the concentration value threshold range, the corresponding offline courses are recommended to the user based on the explicit data; when the concentration value is within the concentration value threshold range, the corresponding offline courses are recommended to the user based on the knowledge level.

[0062] The process of collecting explicit, implicit, and contextual data includes:

[0063] The explicit data includes the user's interest tags (such as mathematics, English, politics, etc., where each interest tag represents a subject, e.g., mathematics is a subject, English is also a subject, and each subject has corresponding online and offline courses). The implicit data includes the user's online course viewing time, online course comment data, and the accuracy rate of the corresponding exercises in the online courses. The contextual data includes the user's offline video recording data and the scores of the corresponding exercises in the offline courses.

[0064] When users register on the intelligent teaching platform, they fill in interest tags, which are stored on the intelligent teaching platform's server. By accessing the intelligent teaching platform's server, we can obtain the user's interest tags, the user's online course viewing time, online course comment data, the accuracy rate of the corresponding exercises in the online course, and the scores of the corresponding exercises in the offline course;

[0065] The video recording data of users’ offline learning is obtained through the camera.

[0066] The process of obtaining users' long-term interest in online courses based on implicit data includes:

[0067] Extract the online course viewing time, online course comment data, and online course corresponding exercise accuracy of users during online learning within T days from the current time, use a two-way matching algorithm to segment the comment data within T days from the current time, collect all the segmented words to obtain a long-term two-way comment word set, set a diverse word database, and store p_v words in the diverse word database, all of which are words in the dictionary. The diverse word database is provided with a positive word storage space, a medium word storage space, and a negative word storage space, which respectively store positive evaluation words (such as good, great, perfect, etc.), medium evaluation words (such as average, okay, so-so, etc.), and negative evaluation words (such as bad, not good, terrible, etc.);

[0068] Compare the words in the long-term two-way review word set with the words stored in the diverse word database. If the word is the same as any positively evaluated word stored in the positive word storage space, then record the word as a positive word. If the word is the same as any medium-evaluated word stored in the medium word storage space, then record the word as a medium word. If the word is the same as any negatively evaluated word stored in the negative word storage space, then record the word as a negative word.

[0069] Obtain the number of positive words, the number of medium words, and the number of negative words in the long-term two-way review word set, and integrate the number of positive words, the number of medium words, and the number of negative words in the long-term two-way review word set through adaptive adjustment to obtain the long-term review impact factor CP i , specifically including:

[0070] in, i is the index of online courses, CP i is the long-term review impact factor corresponding to the i-th online course, CJ i is the number of positive words in the long-term two-way review word set corresponding to the i-th online course, CZ iis the number of medium words in the long-term two-way review word set corresponding to the i-th online course, CX i is the number of negative words in the long-term two-way review word set corresponding to the i-th online course, α is the adjustment factor for the number of positive words in the long-term two-way review word set corresponding to the i-th online course, and α>1. The adjustment factor can be set through experimental data analysis or experience;

[0071] It should be explained that the more positive and medium words there are, the greater the long-term review impact factor is, the more satisfied the user is with the online course, and the fewer negative words there are, The larger it is, the greater the long-term review impact factor is, and the more satisfied the user is with the online course;

[0072] According to the online course viewing time, long-term comment impact factor and the correct rate of the corresponding exercises of the online course during the user's online learning within T days from the current time, the improved weighted average summation method is used to obtain the user's long-term interest value L for the online course i , specifically including:

[0073] in, L i is the user's long-term interest value in the i-th online course, t is the index of T, r it is the viewing time of the i-th online course t days from the current time. For example, the current time is May 12, 2024, t = 2, r i2 is the viewing time of the i-th online course 2 days from the current time, that is, the viewing time of the i-th online course on May 10, 2024, i is the correct rate of the exercises corresponding to the i-th online course, β is the weight coefficient of the viewing time of the i-th online course at t days from the current time, γ is the weight coefficient of the long-term review impact factor corresponding to the i-th online course, and δ is the weight coefficient of the correct rate of the exercises corresponding to the i-th online course. Since the user's interest will change slowly over time, the early learning behavior should be given a lower weight. Therefore, the time decay weighting coefficient is introduced to distribute the weights. it is the time decay weighting coefficient of the i-th online course t days from the current time, w it =e ε(T-t) ; e is the base of the natural logarithm, ε is the decay rate control factor, which is used to control the decay rate. The adjustment coefficient and decay rate control factor can be set through experimental data analysis or experience;

[0074] It should be noted that the long-term interest value of users in online courses is jointly determined by the viewing duration of online courses, the long-term comment influence factor, and the correct rate of answering questions corresponding to the online courses. The traditional method is to perform weighted average summation on the viewing duration of online courses, the long-term comment influence factor, and the correct rate of answering questions corresponding to the online courses to obtain the long-term interest value. The weights of each factor in the long-term interest value do not change, and the influence of time on the change of user interest is not considered, resulting in inaccurate long-term interest values obtained. In this invention, by introducing a time decay weighting coefficient to adjust the weights of the factors, different learning behaviors in different time periods have different weights, so as to capture the change of user interest, ensure the accuracy of the long-term interest value, and then improve the accuracy of resource recommendation on the teaching platform. The longer the viewing duration of the online course, the greater the long-term interest value, and the higher the long-term interest of users in the online course. The greater the long-term comment influence factor, the more satisfied users are with the online course, and the higher the long-term interest of users in the online course. The higher the correct rate of answering questions corresponding to the online course, the more seriously users learn the online course, and further reflects the higher long-term interest of users in the online course;

[0075] The process of obtaining the short-term interest value of users in online courses according to implicit data includes:

[0076] Use a bidirectional matching algorithm to segment the comment data within N days from the current time, N < T, and gather all the words obtained by word segmentation to obtain a short-term bidirectional comment word set;

[0077] For example, T can be 180 and N can be 7. The learning data within 180 days from the current time can reflect the stable learning preferences of users, while the learning data within 7 days from the current time can reflect the content that users have recently learned. Both have a significant impact on the recommendation of resources on the intelligent teaching platform;

[0078] Compare the words in the short-term bidirectional comment word set with the words stored in the diverse word database. If a word is the same as any positive evaluation word stored in the positive word storage space, mark this word as a positive word. If a word is the same as any medium evaluation word stored in the medium word storage space, mark this word as a medium word. If a word is the same as any negative evaluation word stored in the negative word storage space, mark this word as a negative word;

[0079] Obtain the number of positive words, the number of medium words, and the number of negative words in the short-term bidirectional comment word set, and obtain the short-term comment influence factor DP based on the number of positive words, the number of medium words, and the number of negative words in the short-term bidirectional comment word set i ;

[0080] Among them, DP i is the short-term review impact factor corresponding to the i-th online course, DJ i is the number of positive words in the short-term two-way review word set corresponding to the i-th online course, DZ i is the number of medium words in the short-term two-way review word set corresponding to the i-th online course, and DX is the number of negative words in the short-term two-way review word set corresponding to the i-th online course;

[0081] According to the online course viewing time, short-term comment impact factor and the correct rate of the corresponding exercises of the online course during the N days from the current time, the improved weighted average summation method is used to obtain the user's short-term interest value Y for the online course. i , specifically including:

[0082] in, Y i is the user's short-term interest value for the i-th online course, n is the index of N, r in is the viewing time of the i-th online course n days from the current time;

[0083] The process of obtaining the user's comprehensive interest value for online courses based on the long-term interest value and the short-term interest value includes:

[0084] The user boredom coefficient θ is obtained through the Sigmoid function based on the time when the user watched the online course most recently.

[0085] in, a and b are adjustment parameters, which can be set through experimental data analysis or experience to adjust the user boredom coefficient, t recent The time when the user watched the online course most recently from the current time, t 当 is the current time. For example, if the current time is February 14, 2024, and the user's most recent online course viewing time is February 12, 2024, then t 当 -t recent =2;

[0086] Multiply the user's boredom coefficient by the long-term interest value, calculate the difference between the user's boredom coefficient and 1, multiply the difference between the user's boredom coefficient and 1 by the short-term interest value, and add the two multiplication results to obtain the user's comprehensive interest value LY for the online course. i ;

[0087] Among them, LY i =θ×L i +(1-θ)×Y i LY iis the user's comprehensive interest value for the i-th online course. If the user watches the online course every day, the smaller the user's boredom coefficient is, the greater the impact of the short-term interest value on the comprehensive interest value. If the user has not watched the online course for 1 month, the larger the user's boredom coefficient is, the greater the impact of the long-term interest value on the comprehensive interest value.

[0088] It should be explained that the user's comprehensive interest value for online courses is jointly determined by the user's long-term interest value and short-term interest value. The traditional method combines the long-term interest value and the short-term interest value through the weighted average summation method to obtain the comprehensive interest value. The weights of the long-term interest value and the short-term interest value are fixed and not real-time, which results in inaccurate comprehensive interest value, and thus makes the recommended teaching platform resources less accurate and not targeted. Therefore, the present invention dynamically adjusts the weights of the long-term interest value and the short-term interest value by introducing the user boredom coefficient. If the user has watched online courses every day recently, the smaller the user boredom coefficient, the greater the weight of the short-term interest value. If the user has not watched online courses for a long time, the greater the user boredom coefficient, the greater the weight of the long-term interest value, so that the obtained comprehensive interest value is real-time, ensuring the accuracy of the comprehensive interest value, and thus improving the accuracy of the teaching platform resource recommendation;

[0089] The process of recommending corresponding online courses to users based on their comprehensive interest values ​​includes:

[0090] Set a comprehensive interest value threshold. The comprehensive interest value threshold can be set through experimental data analysis or experience. Compare the user's comprehensive interest value for online courses with the comprehensive interest value threshold. Obtain online courses corresponding to comprehensive interest values ​​greater than or equal to the comprehensive interest value threshold and record them as interested online courses.

[0091] Obtain knowledge points from all online courses and connect them based on the relationships between them. The relationships between knowledge points include prerequisite knowledge relationships, hierarchical relationships, and similar concept relationships. For example, the relationship between linear algebra and matrix operations is a prerequisite knowledge relationship, the relationship between machine learning and deep learning is a hierarchical relationship, and the relationship between probability theory and Bayesian statistics is a similar concept relationship.

[0092] If the association relationship between two knowledge points is a pre-knowledge relationship, the two knowledge points are connected and the pre-knowledge relationship is marked on the connection line. If the association relationship between two knowledge points is a hierarchical relationship, the two knowledge points are connected and the hierarchical relationship is marked on the connection line. If the association relationship between two knowledge points is a similar concept relationship, the two knowledge points are connected and the similar concept relationship is marked on the connection line.

[0093] Obtain the knowledge points connected to the knowledge points in the online courses of interest, record them as knowledge points of interest, and recommend online courses corresponding to the knowledge points of interest to users.

[0094] The process of obtaining the user's concentration value for offline courses and the user's knowledge level value based on contextual data includes:

[0095] The user's concentration value for offline courses is obtained based on the video recording data of the user's offline learning, including:

[0096] Obtain the total number of blinks of the user from the video recording data of the user's offline learning, and use the total number of blinks of the user as the user's concentration value for the offline course;

[0097] Get the user's knowledge level value ZS based on the scores of the corresponding exercises in the offline course j ;

[0098] in, j is the index of offline courses, ZS j represents the user's knowledge level relative to the jth offline course, K is the number of exercises corresponding to the jth offline course, k is the index of the number of exercises corresponding to the jth offline course, R jk is the user's score for the kth exercise corresponding to the jth offline course;

[0099] The process of setting the focus threshold range includes:

[0100] Get the range of normal blink times in one minute (YC0, YC1), and set the focus threshold range (YX j ,YD j );

[0101] Among them, YC0 is the minimum value in the range of normal blink times in one minute, YC1 is the maximum value in the range of normal blink times in one minute, and YX j is the minimum value in the concentration threshold range corresponding to the jth offline course, YX j =SC j ×YC0;SC j is the duration of the jth offline course in minutes, YD j YD is the maximum value in the concentration threshold range corresponding to the jth offline course, j =SC j ×YC1;

[0102] The process of obtaining the user's knowledge level according to the user's knowledge level value includes:

[0103] The knowledge level levels include first level knowledge level, second level knowledge level and third level knowledge level;

[0104] It should be noted that the knowledge level of users at the third-level knowledge level is higher than that of users at the second-level knowledge level, and the knowledge level of users at the second-level knowledge level is higher than that of users at the first-level knowledge level;

[0105] Set the threshold range (ZS0, ZS1) of the knowledge level value, and the threshold range of the knowledge level value can be set through experimental data analysis or experience;

[0106] When ZS j < ZS0, the knowledge level of the user relative to the j-th offline course is the first-level knowledge level;

[0107] When ZS0 ≤ ZS j ≤ ZS1, the knowledge level of the user relative to the j-th offline course is the second-level knowledge level;

[0108] When ZS j > ZS1, the knowledge level of the user relative to the j-th offline course is the third-level knowledge level;

[0109] When the focus value exceeds the threshold range of the focus value, the process of recommending the corresponding offline course to the user according to the explicit data includes:

[0110] When the focus value exceeds the threshold range of the focus value, that is, when YC j < YX j or YC j > YD j When, YC j is the focus value of the user on the j-th offline course. When the user is learning this offline course, if the number of blinks is too low, it indicates that the user may be sleeping or发呆. If the number of blinks is too high, it indicates that the user may be nervous. Both too high and too low numbers of blinks indicate that the user is not focused enough on this offline course. Recommend the offline course corresponding to the user's interest label to the user;

[0111] When the focus value is within the threshold range of the focus value, the process of recommending the corresponding offline course to the user according to the knowledge level includes:

[0112] When the focus value is within the threshold range of the focus value, that is, when YX j ≤ YC j ≤ YD j When, the user has a normal number of blinks when learning this offline course, and the user does not show a drowsy or发呆 state, indicating that the user is interested in this offline course;

[0113] It should be explained that the total number of blinks of a user is used as the user's concentration value for the offline course. The fewer the total number of blinks of a user, the more likely the user is sleeping or daydreaming. The more the total number of blinks of a user, the more likely the user is nervous. Both too few and too many total blinks indicate that the user is not focused enough on the offline course. Therefore, the total number of blinks of a user can well represent the user's concentration value for the offline course. Obtaining the concentration value in this way is simpler and faster, which improves the efficiency of resource recommendation on the teaching platform.

[0114] Offline courses for each subject are divided into difficulty levels, including level 1, level 2, and level 3.

[0115] It should be explained that the difficulty of the offline course of level 3 difficulty is greater than the difficulty of the offline course of level 2 difficulty is greater than the difficulty of the offline course of level 1 difficulty;

[0116] When the user's knowledge level relative to the j-th offline course is at level one, obtain the subject corresponding to the j-th offline course and record it as a level one subject. Then recommend the level one offline course corresponding to the level one subject to the user.

[0117] When the user's knowledge level relative to the j-th offline course is at the second-level knowledge level, obtain the subject corresponding to the j-th offline course and record it as a second-level subject. Then recommend the second-level difficulty offline course corresponding to the second-level subject to the user.

[0118] When the user's knowledge level relative to the j-th offline course is at level three, the subject corresponding to the j-th offline course is obtained and recorded as a level three subject, and the offline course of level three difficulty corresponding to the level three subject is recommended to the user.

[0119] All the above formulas are dimensionless and numerically calculated. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0120] In this embodiment, the long-term interest value and short-term interest value of the user in the online course are obtained based on implicit data. In the process of obtaining the long-term interest value, the user's interest will change slowly over time. Early learning behaviors should be given lower weights, and recent learning behaviors should be given higher weights. Therefore, a time decay weighting coefficient is introduced to make the obtained long-term interest value more accurate, which in turn helps to improve the accuracy of resource recommendations on the teaching platform; the user's comprehensive interest value in the online course is obtained based on the long-term interest value and the short-term interest value, and the corresponding online course is recommended to the user based on the comprehensive interest value. For the comprehensive interest value, the user's boredom coefficient is obtained based on the time when the user watched the online course most recently from the current time, and the long-term interest value and the short-term interest value are adjusted based on the user's boredom coefficient to obtain the user's comprehensive interest value in the online course; the user's comprehensive interest is related to the long-term interest and short-term interest. The user's boredom coefficient is obtained based on the time when the user last watched the online course, and the user's boredom coefficient is adjusted based on the user's boredom coefficient. The comprehensive interest value of the user is obtained. If the user watches online courses every day, the smaller the user's boredom coefficient is, the greater the impact of the short-term interest value on the comprehensive interest value is. If the user has not watched online courses for one month, the larger the user's boredom coefficient is, the greater the impact of the long-term interest value on the comprehensive interest value is, so that the obtained comprehensive interest value is more accurate, thereby improving the accuracy of resource recommendations on the teaching platform; according to the contextual data, the user's concentration value for offline courses and the user's knowledge level value are obtained, and the concentration value threshold range is set. According to the user's knowledge level value, the user's knowledge level level is obtained. When the concentration value exceeds the concentration value threshold range, the corresponding offline course is recommended to the user according to the explicit data. When the concentration value is within the concentration value threshold range, the corresponding offline course is recommended to the user according to the knowledge level level; according to the user's knowledge level level, offline courses of corresponding difficulty are recommended to the user, thereby increasing the user's interest in the offline course and increasing the user's motivation to learn, thereby improving the adaptability of resource recommendations on the teaching platform.

[0121] Example 2

[0122] See also Figure 2 As shown, for the parts not described in detail in this embodiment, please refer to the description of Example 1. A resource recommendation management system for an intelligent teaching platform is provided, including:

[0123] Data collection component, responsible for collecting explicit data, implicit data and contextual data;

[0124] The online course recommendation component is responsible for obtaining the user's long-term and short-term interest values ​​for online courses based on implicit data, obtaining the user's comprehensive interest value for online courses based on the long-term and short-term interest values, and recommending corresponding online courses to the user based on the comprehensive interest value;

[0125] The knowledge level acquisition component is responsible for obtaining the user's focus value on offline courses and the user's knowledge level value based on contextual data, setting the focus value threshold range, and obtaining the user's knowledge level based on the user's knowledge level value;

[0126] The offline course recommendation component is responsible for comparing the concentration value with the concentration value threshold range, and recommending corresponding offline courses to users based on the comparison results, explicit data, and knowledge level.

[0127] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0128] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only one type. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0129] The above description is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed by the present invention, which should be covered by the scope of protection of the present invention.

[0130] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A resource recommendation and management method for an intelligent teaching platform, characterized in that: The intelligent teaching platform resource recommendation management method includes: Step S1: Collect explicit data, implicit data and contextual data; Step S2: Obtain the user's long-term interest value and short-term interest value for the online course based on implicit data, obtain the user's comprehensive interest value for the online course based on the long-term interest value and the short-term interest value, and recommend corresponding online courses to the user based on the comprehensive interest value; wherein, for the comprehensive interest value, the user's boredom coefficient is obtained by using the Sigmoid function based on the time when the user watched the online course most recently. Based on the user's boredom coefficient, the long-term interest value and the short-term interest value are fused through adaptive adjustment to obtain the user's comprehensive interest value for the online course; Step S3: Obtain the user's concentration value for the offline course and the user's knowledge level value based on the contextual data, set a concentration value threshold range, and obtain the user's knowledge level level based on the user's knowledge level value; Step S4: When the concentration value exceeds the concentration value threshold range, the corresponding offline courses are recommended to the user based on the explicit data; when the concentration value is within the concentration value threshold range, the corresponding offline courses are recommended to the user based on the knowledge level.

2. The resource recommendation and management method of an intelligent teaching platform according to claim 1, characterized in that: The explicit data includes user interest tags; The implicit data includes the viewing time of online courses, comment data on online courses, and the accuracy rate of exercises corresponding to online courses when users study online. The contextual data includes video recording data of users studying offline and the scores of exercises corresponding to offline courses.

3. The resource recommendation and management method of an intelligent teaching platform according to claim 2, characterized in that: The method for obtaining the user's long-term interest value and short-term interest value for the online course based on implicit data includes: Extract the online course viewing time, online course comment data, and online course corresponding exercise accuracy of users during online learning within T days from the current time, use a two-way matching algorithm to segment the comment data within T days from the current time, collect all the segmented words to obtain a long-term two-way comment word set, set a diverse word database, and store p_v words in the diverse word database. The diverse word database is provided with a positive word storage space, a medium word storage space, and a negative word storage space, which respectively store words with positive evaluation, words with medium evaluation, and words with negative evaluation; Compare the words in the long-term two-way review word set with the words stored in the diverse word database, and obtain the long-term review impact factor based on the comparison result; Based on the online course viewing time, long-term comment impact factor, and correct rate of the corresponding exercises of the online course during the user's online learning within T days from the current time, an improved weighted average summation method is used to obtain the user's long-term interest value in the online course; The user's short-term interest value in the online course is obtained based on the user's online course viewing time, online course comment data, and the accuracy rate of the corresponding exercises in the online course during the N days from the current time.

4. The resource recommendation and management method of an intelligent teaching platform according to claim 3, characterized in that: The method for obtaining the long-term review impact factor according to the comparison result includes: If a word in the long-term two-way comment word set is the same as any positive evaluation word stored in the positive word storage space, then this word is recorded as a positive word. If a word in the long-term two-way comment word set is the same as any medium evaluation word stored in the medium word storage space, then this word is recorded as a medium word. If a word in the long-term two-way comment word set is the same as any negative evaluation word stored in the negative word storage space, then this word is recorded as a negative word; Obtain the number of positive words, the number of medium words, and the number of negative words in the long-term two-way comment word set, and fuse the number of positive words, the number of medium words, and the number of negative words in the long-term two-way comment word set through adaptive adjustment to obtain the long-term comment influence factor.

5. The resource recommendation and management method of an intelligent teaching platform according to claim 4, characterized in that: The method for obtaining the short-term interest value of the user for the online course includes: Use the two-way matching algorithm to segment the comment data within N days from the current time, where N < T, and collect all the words obtained by segmentation to obtain the short-term two-way comment word set; Compare the words in the short-term two-way comment word set with the words stored in the diverse word database, and obtain the short-term comment influence factor according to the comparison result; According to the viewing duration of the online course, the short-term comment influence factor, and the correct rate of doing exercises for the corresponding exercises of the online course during the user's online learning within N days from the current time, use the improved weighted average summation method to obtain the short-term interest value of the user for the online course.

6. The resource recommendation and management method of an intelligent teaching platform according to claim 5, characterized in that: The method for recommending the corresponding online course to the user according to the comprehensive interest value includes: Set the comprehensive interest value threshold, compare the comprehensive interest value of the user for the online course with the comprehensive interest value threshold, obtain the online courses corresponding to the comprehensive interest values greater than or equal to the comprehensive interest value threshold, and record them as interest online courses; Obtain the knowledge points in all online courses, and connect the corresponding knowledge points according to the association relationship between the knowledge points. The association relationship between the knowledge points includes the prerequisite knowledge relationship, the hierarchical relationship, and the similar concept relationship; If the association relationship between two knowledge points is the prerequisite knowledge relationship, then connect the two knowledge points and mark the prerequisite knowledge relationship on the connection line. If the association relationship between two knowledge points is the hierarchical relationship, then connect the two knowledge points and mark the hierarchical relationship on the connection line. If the association relationship between two knowledge points is the similar concept relationship, then connect the two knowledge points and mark the similar concept relationship on the connection line; Obtain the knowledge points connected to the knowledge points in the interest online courses, and record them as interest knowledge points, and recommend the online courses corresponding to the interest knowledge points to the user.

7. The intelligent teaching platform resource recommendation and management method according to claim 6, characterized in that: The method for obtaining the user's concentration value and the user's knowledge level value for the offline course according to the scenario data includes: Obtain the total number of blinks of the user in the video recording data during the user's offline learning, and use the total number of blinks of the user as the user's concentration value for the offline course; Sum up the scores of the exercises corresponding to the offline course to obtain the user's knowledge level value.

8. The intelligent teaching platform resource recommendation and management method according to claim 7, characterized in that: The method for setting the concentration value threshold range includes: Obtain the normal blink count range per minute and the duration of the offline course; Calculate the product of the minimum value in the normal number of blinks in one minute and the offline course duration to obtain the minimum value in the concentration value threshold range. Calculate the product of the maximum value in the normal number of blinks in one minute and the offline course duration to obtain the maximum value in the concentration value threshold range.

9. The resource recommendation and management method of an intelligent teaching platform according to claim 8, characterized in that: The method for obtaining the user's knowledge level grade according to the user's knowledge level value includes: The knowledge level levels include first level knowledge level, second level knowledge level and third level knowledge level; Set the knowledge level value threshold range; When the user's knowledge level value is less than the minimum value in the knowledge level value threshold range, the user's knowledge level relative to the offline course is the first level knowledge level; When the user's knowledge level value is less than or equal to the maximum value in the knowledge level value threshold range and greater than or equal to the minimum value in the knowledge level value threshold range, the user's knowledge level level relative to the offline course is the second level knowledge level; When the user's knowledge level value is greater than the maximum value in the knowledge level value threshold range, the user's knowledge level rank relative to the offline course is the third knowledge level.

10. The intelligent teaching platform resource recommendation and management method according to claim 9, characterized in that: The method of recommending corresponding offline courses to the user based on explicit data when the concentration value exceeds the concentration value threshold range, and recommending corresponding offline courses to the user based on the knowledge level when the concentration value is within the concentration value threshold range includes: When the concentration value exceeds the concentration value threshold range, the offline courses corresponding to the user's interest tags are recommended to the user; When the concentration value is within the concentration value threshold range, the corresponding offline courses are recommended to the user based on the knowledge level; Offline courses for each subject are divided into difficulty levels, including level 1, level 2, and level 3. When the user's knowledge level is at level one relative to the offline course, the subject corresponding to the offline course is obtained and recorded as a level one subject. The offline course of level one difficulty corresponding to the level one subject is recommended to the user. When the user's knowledge level is at level 2 relative to the offline course, the subject corresponding to the offline course is obtained and recorded as a level 2 subject, and offline courses of level 2 difficulty corresponding to the level 2 subject are recommended to the user; When the user's knowledge level relative to the offline course is at level three, the subject corresponding to the offline course is obtained and recorded as a level three subject, and an offline course of level three difficulty corresponding to the level three subject is recommended to the user.

Images