Personalized learning path planning system and method based on intelligent recommendation
By analyzing user learning logs and historical data, and using genetic algorithms and Apriori algorithms to optimize learning paths, the problem of failing to provide personalized learning paths in the existing technology is solved, and effective learning path planning and weak knowledge points are achieved.
Patent Information
- Application Number
- CN202411281301.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-13
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2044-09-13
AI Technical Summary
The existing learning path planning method for online education fails to fully consider students' individual differentiated needs and real-time feedback, making it difficult to provide personalized learning path planning, unable to effectively improve weak knowledge points, and it is difficult to recommend reasonable and in line with users in many courses. Courses of interest.
By obtaining the online learning log and historical test records of the target user, the knowledge points of interest and knowledge points that have not been mastered are extracted, the genetic algorithm and Apriori algorithm are used to filter out the appropriate recommended learning video resources, and the interest and mastery degree measurement model is built. , optimize learning paths to provide personalized learning path planning.
It has achieved personalized learning path planning for students in a diversified curriculum environment to ensure learning effectiveness and maximize the mastery of weak knowledge points.
Smart Images

Figure CN118799142B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of computer technology, and more specifically, to a personalized learning path planning system and method based on intelligent recommendation. Background Art
[0002] With the continuous advancement of educational technology and the popularization of online education platforms and digital learning resources, modern education is no longer limited to traditional classroom teaching, and can enable students to access a rich variety of learning content and course options. Students can independently choose and study various courses according to their own interests and learning goals; but with the increase in the number and types of courses available, students often face the problem of "difficulty in selection" when they manually select learning courses; at the same time, due to the lack of sufficient guidance, suggestions and evaluation, it is difficult to extract more reasonable recommended courses that meet the interests of users from many options; in addition, the existing learning path planning methods for online education are usually limited to existing courses, while ignoring the differentiated needs of individual students, and failing to fully consider the real-time feedback and personalized needs shown by students in the learning process, resulting in difficulty in flexible learning path planning based on course recommendation results. Further, it is difficult to help students obtain the best course recommendation results and plan the best learning path, and thus it is impossible to provide personalized learning path planning services for different students. Furthermore, it is impossible to ensure the effectiveness of students' learning and to help students maximize the improvement of weak knowledge points. Summary of the invention
[0003] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a personalized learning path planning system and method based on intelligent recommendation.
[0004] To achieve the above object, the present invention provides the following technical solutions:
[0005] A personalized learning path planning method based on intelligent recommendation, the method comprising:
[0006] According to the recorded online learning log, obtain the knowledge point set A that the target user has learned, and extract the knowledge point set B that the target user is interested in from the knowledge point set A; the knowledge point set A contains M learned knowledge points, and the knowledge point set B contains N interesting knowledge points, M≥N>0, M, N∈N + , N + is a positive integer;
[0007] Obtain the mastery degree value of each knowledge point of interest for the target user, obtain U mastery degree values, and filter out Q unmastered knowledge points of interest for the target user based on the mastery degree values, U and Q> 0, U and Q∈N + ;
[0008] Acquire multiple non-target users similar to the target user, retrieve historical learning experience data of each non-target user, and screen out recommended learning video resources that are compatible with each unlearned knowledge point of interest of the target user based on the historical learning experience data to obtain a recommended course set; the historical learning experience data includes multiple learned knowledge points of the non-target user and selected watched learning video resources for each learned knowledge point;
[0009] The recommended course set is input into a preconfigured genetic algorithm to plan an optimal learning path for the target user for the recommended course set, wherein the learning path represents a learning order of the recommended learning video resources.
[0010] Furthermore, the step of extracting a set of knowledge points B that the target user is interested in includes:
[0011] a1: Extract the target user's learning behavior information for the i-th learned knowledge point in the knowledge point set A, where the learning behavior information includes the target user's repeated learning times for the i-th learned knowledge point and the learning duration of each repeated learning, i>0, i∈N + ;
[0012] a2: Input the learning behavior information of the i-th learned knowledge point into the pre-built interest measurement model to obtain the target user's interest coefficient for the i-th learned knowledge point;
[0013] The mathematical expression of the pre-built interest measurement model is as follows:
[0014] ;
[0015] Where: is the interest coefficient of the i-th learned knowledge point; Indicates the number of times the target user repeats learning the i-th learned knowledge point; Indicates The learning time of the i-th learned knowledge point; is the time constant; is a natural constant; and are the sine and cosine functions respectively; is a logarithmic function;
[0016] a3: Compare the interest coefficient of the i-th learned knowledge point with the preset interest coefficient threshold. If the interest coefficient is greater than or equal to the interest coefficient threshold, mark the i-th learned knowledge point as an interest knowledge point, set i=i+1, and return to step a1; if the interest coefficient is less than the interest coefficient threshold, set i=i+1 and directly return to step a1;
[0017] a4: Repeat the above steps a1 to a3 until i=M, then end the loop, obtain N knowledge points of interest, and merge the N knowledge points of interest into a knowledge point set B.
[0018] Furthermore, obtaining the target user's mastery level value for each knowledge point of interest includes:
[0019] b1: Extract the jth interest knowledge point in the knowledge point set B, and retrieve the target user's assessment data for the jth interest knowledge point based on the historical test records, j>0, j∈N + ;
[0020] The assessment data includes the assessment scores, assessment time and assessment error rate of the target user for the learned knowledge points in each special assessment;
[0021] b2: Input the assessment data of the jth interest knowledge point into the pre-built mastery degree measurement model to obtain the mastery degree value of the target user for the jth interest knowledge point; set j=j+1, and return to step b1;
[0022] The mathematical expression of the pre-built mastery degree measurement model is as follows:
[0023] ;
[0024] Where: is the mastery level value of the jth interest knowledge point; is the total number of special assessments for the target user on the jth knowledge point of interest; is the target user's assessment score for the jth interest knowledge point in the vth special assessment; is the target user's error rate for the jth interest knowledge point in the vth special assessment; The time taken by the target user to assess the jth knowledge point of interest in the vth special assessment; is the time correction factor;
[0025] b3: Repeat the above steps b1 to b2 until j=N, then end the loop and obtain the target user's mastery level value for each knowledge point of interest.
[0026] Furthermore, the step of screening out Q unlearned knowledge points of interest to the target user includes:
[0027] c1: Normalize the mastery level of the gth interest knowledge point in the knowledge point set B, g>0, g∈N + ;
[0028] c2: Compare the normalized mastery value of the g-th interest knowledge point with the preset mastery threshold. If the mastery value is greater than the mastery threshold, mark the g-th interest knowledge point as a mastered interest knowledge point, set g=g+1, and return to step c1; if the mastery value is less than or equal to the mastery threshold, mark the g-th interest knowledge point as an unmastered interest knowledge point, set g=g+1, and return to step c1;
[0029] c3: Repeat the above steps c1 to c2 until g=M, then end the loop and obtain Q unmastered knowledge points of interest.
[0030] Furthermore, the step of screening out recommended learning video resources that are compatible with each of the target user's unlearned knowledge points of interest includes:
[0031] d1: extracting a knowledge point set C of non-target users from a plurality of learned knowledge points of non-target users; the knowledge point set C includes a plurality of knowledge points of interest that the non-target users have mastered;
[0032] d2: Extract the acquired knowledge points in the knowledge point set C that are the same as the target user's Lth unacquired knowledge point, and mark them as overlapping knowledge points, L>0, L∈N + ;
[0033] d3: Based on the learning video resources that non-target users choose to watch for each learned knowledge point, extract the learning video resources that are selected to watch for the overlapping knowledge points and mark them as candidate learning video resources;
[0034] d4: Use the Apriori algorithm to calculate the support of each candidate learning video resource, and sort the candidate learning video resource with the highest support from the largest to the smallest value as the recommended learning video resource that matches the target user's Lth unmastered interest point, and set L=L+1, and return to step d2;
[0035] d5: Repeat steps d2 to d4 until L=Q, then end the loop and obtain recommended learning video resources that match each of the target user's unmastered knowledge points of interest. Merge the recommended learning video resources that match each of the target user's unmastered knowledge points of interest into a recommended course set.
[0036] Furthermore, the logic for obtaining the optimal learning path is as follows:
[0037] e1: Initialize the population: randomly generate an original population, which contains X chromosomes, each chromosome represents a learning path, X>0, X∈N + ;
[0038] e2: Fitness evaluation: Under each chromosome (i.e., each learning path in the original population), obtain the target user's mastery gain value for each unmastered knowledge point of interest, input the mastery gain value of each unmastered knowledge point of interest into the pre-built fitness function, and calculate the fitness of each chromosome;
[0039] e3: Selection: Use the roulette wheel method to select two chromosomes with high fitness in the original population as the father and mother;
[0040] e4: Crossover: Perform a crossover operation on the father and mother to generate new chromosomes;
[0041] e5: Mutation: Perform mutation operation on the new chromosomes to obtain Y new chromosomes, combine the Y new chromosomes into a new population, replace the original population with the new population, and return to step e2;
[0042] e6: Repeat steps e2 to e5 above until the fitness of the chromosome in the original population or the new population is greater than or equal to the preset fitness threshold, or the number of iterations is greater than or equal to the preset maximum number of iterations threshold, and output the learning path represented by the corresponding chromosome as the optimal learning path.
[0043] Furthermore, the mastery degree gain value is calculated according to a predefined gain function, and its specific calculation formula is as follows:
[0044] ;
[0045] Where: represents the mastery gain value of the jth unmastered interest knowledge point, Indicates the degree of mastery after viewing. Indicates the initial mastery value, that is, the mastery value before watching.
[0046] Furthermore, the calculation formula of the pre-constructed fitness function is as follows: ; Where: For fitness; is the total number of knowledge points of interest that have not been mastered; is the median function;
[0047] The specific expression of the median function is as follows:
[0048] ;
[0049] Where: Represents the total number of gain values of the mastery degree of the j-th knowledge point of interest that has not been mastered.
[0050] A personalized learning path planning system based on intelligent recommendation, comprising:
[0051] The first screening module is used to obtain the knowledge point set A that the target user has learned based on the recorded online learning log, and extract the knowledge point set B that the target user is interested in from the knowledge point set A; the knowledge point set A contains M learned knowledge points, and the knowledge point set B contains N interesting knowledge points, M≥N>0, M, N∈N + , N + is a positive integer;
[0052] The second screening module is used to obtain the mastery degree value of each interest knowledge point of the target user, obtain U mastery degree values, and screen out Q unmastered interest knowledge points for the target user based on the mastery degree values, U and Q> 0, U and Q∈N + ;
[0053] The course recommendation module is used to obtain multiple non-target users similar to the target user, and retrieve the historical learning experience data of each non-target user, and screen out recommended learning video resources that are compatible with each unlearned knowledge point of interest of the target user according to the historical learning experience data, to obtain a recommended course set; the historical learning experience data includes multiple learned knowledge points of the non-target user and the learning video resources that have been selected and watched for each learned knowledge point;
[0054] The path optimization module is used to input the recommended course set into a preconfigured genetic algorithm to plan the optimal learning path for the target user for the recommended course set, and the learning path represents the learning order of the recommended learning video resources.
[0055] Compared with the prior art, the present invention has the following beneficial effects:
[0056] The present application discloses a personalized learning path planning system and method based on intelligent recommendation, comprising: extracting a knowledge point set B of interest to a target user from a knowledge point set A; screening out Q unmastered knowledge points of interest to the target user based on a mastery degree value; retrieving historical learning experience data of each non-target user, and screening out recommended learning video resources that are compatible with each unmastered knowledge point of interest of the target user according to the historical learning experience data, to obtain a recommended course set; inputting the recommended course set into a preconfigured genetic algorithm to plan out an optimal learning path for the target user with respect to the recommended course set; based on the above technical features, the present invention can match the target user with reasonable recommended courses that are in line with the user's interests when the number and types of courses available for selection are increasing. Furthermore, on the basis of obtaining the recommended courses, the recommended courses are flexibly planned for learning paths through an optimization algorithm. The present invention can provide personalized learning path planning services for different students, and further, is conducive to ensuring the effectiveness of student learning and maximizing the consolidation and improvement of weak knowledge points. BRIEF DESCRIPTION OF THE DRAWINGS
[0057] Figure 1 A flowchart of a personalized learning path planning method based on intelligent recommendation provided by the present invention;
[0058] Figure 2 A schematic diagram of the module structure of a personalized learning path planning system based on intelligent recommendation provided by the present invention. DETAILED DESCRIPTION
[0059] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.
[0060] Example 1
[0061] See also Figure 1 As shown, this embodiment discloses a personalized learning path planning method based on intelligent recommendation, and the method includes:
[0062] S101: Based on the recorded online learning log, obtain a set A of knowledge points that the target user has learned, and extract a set B of knowledge points that the target user is interested in from the set A of knowledge points; the set A of knowledge points contains M learned knowledge points, and the set B of knowledge points contains N knowledge points of interest, M≥N>0, M, N∈N + , N + is a positive integer;
[0063] It should be understood that: a background monitoring tool is usually set up in the online education platform to monitor the learning process (including but not limited to information such as learned knowledge points) and test records of each target user, and record them in the form of logs to form online learning logs and historical test records, which are stored in the server of the online education platform;
[0064] In implementation, the step of extracting a set of knowledge points B that the target user is interested in includes:
[0065] a1: Extract the target user's learning behavior information for the i-th learned knowledge point in the knowledge point set A, where the learning behavior information includes the target user's repeated learning times for the i-th learned knowledge point and the learning duration of each repeated learning, i>0, i∈N + ;
[0066] It should be understood that: the learning behavior information of each learned knowledge point in the knowledge point set A is recorded in the online learning log;
[0067] a2: Input the learning behavior information of the i-th learned knowledge point into the pre-built interest measurement model to obtain the target user's interest coefficient for the i-th learned knowledge point;
[0068] The mathematical expression of the pre-built interest measurement model is as follows:
[0069] ;
[0070] Where: is the interest coefficient of the i-th learned knowledge point; Indicates the number of times the target user repeats learning the i-th learned knowledge point; Indicates The learning time of the i-th learned knowledge point; is a time constant used to simulate the learning duration the attenuating or enhancing effect of the change on interest; is a natural constant; and They are sine and cosine functions, used to capture learning time The impact of periodic changes on interest; Is a logarithmic function used to adjust the number of repeated learning Nonlinear effects on interest;
[0071] a3: Compare the interest coefficient of the i-th learned knowledge point with the preset interest coefficient threshold. If the interest coefficient is greater than or equal to the interest coefficient threshold, mark the i-th learned knowledge point as an interest knowledge point, set i=i+1, and return to step a1; if the interest coefficient is less than the interest coefficient threshold, set i=i+1 and directly return to step a1;
[0072] It can be understood that: when the interest coefficient is less than the interest coefficient threshold, it means that the target user has a low interest or no interest in the i-th learned knowledge point; but it should be understood that: this only means that the target user has a low interest or no interest in the learned knowledge point at the current moment; but with the change of time and the change of user behavior, the interest coefficient of this learned knowledge point may gradually increase and form an interesting knowledge point;
[0073] a4: Repeat the above steps a1 to a3 until i=M, then end the loop, obtain N knowledge points of interest, and merge the N knowledge points of interest into a knowledge point set B;
[0074] It is understandable that: when the target user repeatedly studies a certain learned knowledge point for a long time, it indicates that the target user attaches great importance to the learned knowledge point, and it can be regarded as the target user's interest knowledge point.
[0075] S102: Obtain the mastery level of each knowledge point of interest of the target user, obtain U mastery level values, and filter out Q knowledge points of interest that the target user has not mastered based on the mastery level values, where U and Q>0, U and Q∈N + ;
[0076] In implementation, obtaining the target user's mastery level value for each knowledge point of interest includes:
[0077] b1: Extract the jth interest knowledge point in the knowledge point set B, and retrieve the target user's assessment data for the jth interest knowledge point based on the historical test records, j>0, j∈N + ;
[0078] The historical test records include the target user's assessment data for each learned knowledge point, and the assessment data includes the target user's assessment score, assessment time and assessment error rate for each special assessment.
[0079] b2: Input the assessment data of the jth interest knowledge point into the pre-built mastery degree measurement model to obtain the mastery degree value of the target user for the jth interest knowledge point; set j=j+1, and return to step b1;
[0080] The mathematical expression of the pre-built mastery degree measurement model is as follows:
[0081] ;
[0082] Where: is the mastery level value of the jth interest knowledge point; is the total number of special assessments for the target user on the jth knowledge point of interest; is the target user's assessment score for the jth interest knowledge point in the vth special assessment; is the target user's error rate for the jth interest knowledge point in the vth special assessment; The time taken by the target user to assess the jth knowledge point of interest in the vth special assessment; is the time correction factor;
[0083] b3: Repeat the above steps b1 to b2 until j=N, then end the loop and obtain the target user's mastery level value for each knowledge point of interest, i.e., U mastery level values;
[0084] In implementation, the step of screening out Q unlearned knowledge points of interest to the target user includes:
[0085] c1: Normalize the mastery level of the gth interest knowledge point in the knowledge point set B, g>0, g∈N + ;
[0086] It should be noted that: after obtaining the target user's mastery of the jth interest knowledge point, it will be normalized for the convenience of calculation. The specific normalization formula is: ; Where: is the mastery level value of the jth interest knowledge point after normalization; It is the minimum mastery level value among all the knowledge points of interest; The maximum mastery value among all the knowledge points of interest;
[0087] c2: Compare the normalized mastery value of the g-th interest knowledge point with the preset mastery threshold. If the mastery value is greater than the mastery threshold, mark the g-th interest knowledge point as a mastered interest knowledge point, set g=g+1, and return to step c1; if the mastery value is less than or equal to the mastery threshold, mark the g-th interest knowledge point as an unmastered interest knowledge point, set g=g+1, and return to step c1;
[0088] It should be noted that: if the g-th interest knowledge point is marked as a mastered interest knowledge point, it means that the target user is very familiar with and understands the interest knowledge point, so that when answering questions related to the interest knowledge point, the correct rate is very high; correspondingly, if the g-th interest knowledge point is marked as an unmastered interest knowledge point, it means that when answering questions related to the interest knowledge point, the target user has a high error rate, which further indicates that the target user has not mastered and is not familiar with the interest knowledge point well;
[0089] c3: Repeat the above steps c1 to c2 until g=M, then end the loop and obtain Q unlearned knowledge points of interest;
[0090] For example, it is assumed that there is a target user who has two knowledge points of interest, namely K1 (“linear equation of one variable”) and K2 (“linear equation of two variables”), and has been assessed twice on K1 and K2 respectively. The specific assessment data is shown in Table 1 below:
[0091] Table 1: Special assessment data table
[0092] First assessment score First test error rate First assessment time Second assessment score Second test error rate Second assessment time K1 70 0.2 35 80 0.1 40 K2 60 0.3 40 75 0.25 35
[0093] If the time correction factor is 0.1, and then the data in Table 1 are imported into the mastery degree measurement model for calculation and normalization. It is obtained that the normalized mastery degree value of K1 is greater than 0.6, and the normalized mastery degree value of K2 is less than 0.4. Assuming that the preset mastery degree threshold is 0.5, it is determined that K1 (which is "a linear equation of one variable") is the target user's knowledge point of interest that has been mastered, and K2 (which is "a linear equation of two variables") is the target user's knowledge point of interest that has not been mastered.
[0094] S103: Acquire multiple non-target users similar to the target user, retrieve historical learning experience data of each non-target user, and screen out recommended learning video resources that are compatible with each unlearned knowledge point of interest of the target user according to the historical learning experience data to obtain a recommended course set; the historical learning experience data includes multiple learned knowledge points of the non-target user and selected watched learning video resources for each learned knowledge point;
[0095] It should be noted that: the multiple non-target users similar to the target user can be obtained after screening according to predefined similarity screening conditions, or after screening according to a similarity algorithm (such as a cosine similarity algorithm) or a clustering algorithm (such as a K-means algorithm), and the predefined similarity screening conditions include but are not limited to user age, grade of study, and learning progress, etc.;
[0096] For example, assuming that the target user is 12 years old, is in the sixth grade of primary school, and has a learning progress of 70 (assuming that a course has 10 units and the student has completed 7 units, then the course completion rate is 70%); and it is found on the online education platform that there are 5 non-target users who are completely consistent with the predefined similarity screening conditions, then these 5 non-target users are regarded as non-target users similar to the target user;
[0097] In implementation, the method of screening out recommended learning video resources that are compatible with each unlearned knowledge point of interest of the target user includes:
[0098] d1: extracting a knowledge point set C of non-target users from a plurality of learned knowledge points of non-target users; the knowledge point set C includes a plurality of knowledge points of interest that the non-target users have mastered;
[0099] d2: Extract the acquired knowledge points in the knowledge point set C that are the same as the target user's Lth unacquired knowledge point, and mark them as overlapping knowledge points, L>0, L∈N + ;
[0100] d3: Based on the learning video resources that non-target users choose to watch for each learned knowledge point, extract the learning video resources that are selected to watch for the overlapping knowledge points and mark them as candidate learning video resources;
[0101] d4: Use the Apriori algorithm to calculate the support of each candidate learning video resource, and sort the candidate learning video resource with the highest support from the largest to the smallest value as the recommended learning video resource that matches the target user's Lth unmastered interest point, and set L=L+1, and return to step d2;
[0102] d5: Repeat the above steps d2 to d4 until L=Q, then end the loop, obtain the recommended learning video resources that match each of the target user's unmastered knowledge points of interest, and merge the recommended learning video resources that match each of the target user's unmastered knowledge points of interest into a recommended course set;
[0103] For example, it is assumed that the target user has two knowledge points of interest that he has not mastered, and it is also assumed that through searching, it is found that there are 5 non-target users who are similar to the target user, and the 5 non-target users have 5 knowledge points that they have learned, as shown in Table 2 below:
[0104] Table 2: Data table of knowledge points learned by non-target users and target users
[0105] Mastery value user Knowledge Point A Knowledge Point B Knowledge Point C (Overlapping) Knowledge Point D (Overlapping) Knowledge Point E Target Users / / / 0.21 0.4 Non-target user 1 0.91 0.72 0.43 0.61 0.89 Non-target user 2 0.67 0.94 0.32 0.78 0.86 Non-target users 3 0.75 0.89 0.90 0.96 0.26 Non-target users 4 0.63 0.54 0.25 0.85 0.56 Non-target users 5 0.88 0.78 0.65 0.59 0.63
[0106] By observing Table 2 above, it is found that the five non-target users have five learned knowledge points (Knowledge Point A, Knowledge Point B, Knowledge Point C, Knowledge Point D and Knowledge Point E), while the target user has only two learned knowledge points (Knowledge Point D and Knowledge Point E). According to the relevant example content of the position in Table 1 above, the target user only has two learned knowledge points (Knowledge Point D and Knowledge Point E), and the mastery values of these two learned knowledge points are both lower than the preset mastery threshold (0.5). Therefore, Knowledge Point D and Knowledge Point E are the unmastered interests of the target user. knowledge point; for knowledge point D, the mastery degree values of non-target users 1-5 are all greater than 0.5, which means that for non-target users 1-5, knowledge point D belongs to the same knowledge point of interest that the target user has not mastered, that is, it is an overlapping knowledge point; and for knowledge point E, since only non-target user 3 has a mastery degree value less than 0.5, it means that for non-target users 1, 2, 4, and 5, knowledge point E belongs to the same knowledge point of interest that the target user has not mastered, that is, it is an overlapping knowledge point;
[0107] Continuing with the above example, assume that for the overlapping knowledge point (knowledge point D), the historical learning experience data records the learning video resources that 5 non-target users chose to watch. For details, see Table 3 below:
[0108] Table 3: Data table of learning video resources selected by non-target users for knowledge point D
[0109] user Viewing history (for knowledge point D) Non-target user 1 Video V11, Video V12 Non-target user 2 Video V11, Video V13 Non-target users 3 Video V12, Video V13, Video V14 Non-target users 4 Video V11, Video V12, Video V13 Non-target users 5 Video V11, Video V12, Video V14
[0110] From Table 3 above, we can see that for knowledge point D, the learning video resources selected by the five non-target users are not the same. In general, they include video V11, video V12, video V13 and video D. Since video V11, video V12, video V13 and video D are learning video resources selected for the overlapping knowledge point (knowledge point D), they are marked as candidate learning video resources. After using the Apriori algorithm to calculate the support of the above candidate learning video resources, it is found that the support of video V11 (support = 4 / 5 = 0.8) is ranked first, that is, the support of video V11 is the largest. Therefore, it means that watching video V11 is the most helpful for the learning improvement of the overlapping knowledge point (knowledge point D), and then video V11 is used as a recommended learning video resource that is suitable for the target user's knowledge point of interest that has not been mastered. Similarly, for the overlapping knowledge point (knowledge point E), the same is true, and it will not be repeated here.
[0111] It is worth noting that: if video V11 has been watched by the target user, the candidate learning video resource ranked second according to the extracted support will be used as the recommended learning video resource that is suitable for the knowledge points of interest that the target user has not mastered, and so on; but it should be understood that: with the continuous enrichment of the platform's learning video resources and the increase in similar non-target users, the recommended learning video resources found will not have very low support.
[0112] S104: inputting the recommended course set into a preconfigured genetic algorithm to plan an optimal learning path for the target user for the recommended course set, wherein the learning path represents a learning order of the recommended learning video resources;
[0113] Specifically, the logic for obtaining the optimal learning path is as follows:
[0114] e1: Initialize the population: randomly generate an original population, which contains X chromosomes, each chromosome represents a learning path, X>0, X∈N + ;
[0115] For example, following and referring to the above-mentioned related example content, it is assumed that the recommended course set includes recommended learning video resources V11 and V21, wherein V11 is a recommended learning video resource for knowledge point D, and V21 is a recommended learning video resource for knowledge point E; then there are two learning paths, namely V11→V21 and V21→V11;
[0116] e2: Fitness evaluation: Under each chromosome (i.e., each learning path in the original population), obtain the target user's mastery gain value for each unmastered knowledge point of interest, input the mastery gain value of each unmastered knowledge point of interest into the pre-built fitness function, and calculate the fitness of each chromosome;
[0117] The mastery degree gain value is calculated according to a predefined gain function, and its specific calculation formula is as follows:
[0118] ;
[0119] Where: represents the mastery gain value of the jth unmastered interest knowledge point, Indicates the degree of mastery after viewing. Indicates the initial mastery value, that is, the mastery value before watching;
[0120] It should be noted that: and The calculation logic is the same as above The calculation formula is the same as that of The difference is that is the mastery level of non-target users (similar to target users) before watching the recommended learning video resources, and It is the mastery level of non-target users (similar to target users) after choosing a certain learning path (i.e. learning order) and watching the recommended learning video resources;
[0121] The calculation formula of the pre-constructed fitness function is as follows: ; Where: For fitness; is the total number of knowledge points of interest that have not been mastered; is the median function;
[0122] The specific expression of the median function is as follows:
[0123] ;
[0124] Where: Indicates the total number of mastery gain values of the jth knowledge point of interest that has not been mastered;
[0125] Exemplarily, following the above-mentioned related example content, it is assumed that it is found in the online platform that there are 5 non-target users who are similar to the target user, and have 2 identical overlapping knowledge points with the target user, namely knowledge point D and knowledge point E, and after screening for knowledge point D and knowledge point E according to the above content, a recommended course set is obtained, and the recommended course set includes recommended learning video resources V11 and V21; among them, non-target users 1-3 choose to watch knowledge point D and knowledge point E with the learning path of V11→V21, and non-target users 4-5 choose to watch knowledge point D and knowledge point E with the learning path of V21→V11. According to the selection of different learning paths, their mastery gain values are respectively collected. For details, see the following Tables 4 and 5:
[0126] Table 4: Data table of mastery gain when choosing different learning paths for knowledge point D
[0127] Learning Path (Sequence of Learning) Initial mastery level value Mastery value after viewing Mastery gain value Non-target user 1 V11 → V21 0.21 0.61 0.40 Non-target user 2 V11 → V21 0.40 0.78 0.38 Non-target users 3 V11 → V21 0.49 0.96 0.47 Non-target users 4 V21 → V11 0.46 0.85 0.39 Non-target users 5 V21 → V11 0.38 0.59 0.21
[0128] Table 5: Data table of mastery gain when choosing different learning paths for knowledge point E
[0129] Learning Path (Sequence of Learning) Initial mastery value Mastery value after viewing Mastery gain value Non-target user 1 V11 → V21 0.49 0.89 0.30 Non-target user 2 V11 → V21 0.34 0.86 0.52 Non-target users 4 V21 → V11 0.42 0.56 0.14 Non-target users 5 V21 → V11 0.47 0.63 0.16
[0130] It is worth noting that the mastery level value of non-target user 3 for knowledge point E does not reach the mastery level threshold, so it can be regarded as data that does not meet the calculation conditions or an abnormal value, and is eliminated when calculating the mastery level gain value of non-target user 3 for knowledge point E;
[0131] It can be seen from Tables 4 and 5 above that non-target users 1-3 choose to watch knowledge points D and E using the learning path of V11→V21, and their data are substituted into the fitness function to obtain: ; For knowledge points D and E, non-target users 4-5 choose to watch them with the learning path of V21→V11, and their data are substituted into the fitness function to obtain: ; By comparing the fitness functions of the two learning paths, it is found that the fitness value of the V11→V21 learning path is greater than the fitness value of the V21→V11 learning path. Therefore, V11→V21 can be regarded as the optimal learning path;
[0132] It is worth noting that due to the limitation of the size of the recommended course set, only two learning paths appear in the above example. However, in actual application, new learning paths will be generated and iterated according to the genetic algorithm mechanism until the iteration end condition is met, and then the optimal learning path will be obtained.
[0133] e3: Selection: Use the roulette wheel method to select two chromosomes with high fitness in the original population as the father and mother;
[0134] The roulette wheel method is a commonly used selection method. It is used in genetic algorithms to select chromosomes with higher fitness to enter the next generation. It simulates the roulette wheel process. Each chromosome obtains a corresponding "roulette wheel" area according to its fitness. The higher the fitness of the chromosome, the larger its corresponding area, and the higher the probability of being selected.
[0135] e4: Crossover: Perform a crossover operation on the father and mother to generate new chromosomes;
[0136] It should be noted that the crossover operation on the paternal parent and the maternal parent is implemented based on a crossover operation, and the crossover operation includes but is not limited to one of a single-point crossover, a uniform crossover or a sequential crossover, etc.;
[0137] e5: Mutation: Perform mutation operation on the new chromosomes to obtain Y new chromosomes, combine the Y new chromosomes into a new population, replace the original population with the new population, and return to step e2;
[0138] In the genetic algorithm, the mutation operation is to introduce gene diversity and prevent the algorithm from falling into the local optimum; the mutation operation on the new chromosome is implemented by uniform mutation or Gaussian mutation;
[0139] e6: Repeat steps e2 to e5 above until the fitness of the chromosome in the original population or the new population is greater than or equal to the preset fitness threshold, or the number of iterations is greater than or equal to the preset maximum number of iterations threshold, and output the learning path represented by the corresponding chromosome as the optimal learning path;
[0140] For example, assuming that the maximum number of iterations is 100, the chromosome with the highest fitness in the current population and its fitness value are recorded after each iteration; if it is found that the fitness value has not changed significantly in a certain generation, it is considered that the convergence condition is met, the iteration is stopped, and the learning path represented by the corresponding chromosome is output as the optimal learning path;
[0141] It should be noted that: after obtaining the recommended course set and the optimal learning path for the recommended course set, they will be fed back to the target users to encourage them to conduct intensive learning on the knowledge points of interest that they have not mastered, so as to enable the target users to consolidate and improve the knowledge points that they are interested in but weak in, so as to encourage them to improve their familiarity and understanding of the knowledge points of interest that they have not mastered.
[0142] Example 2
[0143] See also Figure 2As shown, based on the same inventive concept, this embodiment discloses a personalized learning path planning system based on intelligent recommendation. For details not provided in this embodiment, please refer to the description of the relevant parts in Embodiment 1. The system includes:
[0144] The first screening module 210 is used to obtain a set A of knowledge points that the target user has learned based on the recorded online learning log, and extract a set B of knowledge points that the target user is interested in from the set A of knowledge points; the set A of knowledge points contains M learned knowledge points, and the set B of knowledge points contains N knowledge points of interest, M≥N>0, M, N∈N + , N + is a positive integer;
[0145] The second screening module 220 is used to obtain the mastery level value of each interest knowledge point of the target user, obtain U mastery level values, and screen out Q unmastered interest knowledge points for the target user based on the mastery level values, where U and Q>0, U and Q∈N + ;
[0146] The course recommendation module 230 is used to obtain multiple non-target users similar to the target user, and retrieve the historical learning experience data of each non-target user, and filter out recommended learning video resources that are compatible with each unlearned knowledge point of interest of the target user according to the historical learning experience data, to obtain a recommended course set; the historical learning experience data includes multiple learned knowledge points of the non-target user and the learning video resources that have been selected and watched for each learned knowledge point;
[0147] The path optimization module 240 is used to input the recommended course set into a preconfigured genetic algorithm to plan the optimal learning path for the target user for the recommended course set, wherein the learning path represents the learning order of the recommended learning video resources.
[0148] The above formulas are all dimensionless and numerical calculations. 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, weights and thresholds in the formula are set by technicians in this field according to actual conditions.
[0149] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website, computer, server or data center to another website, computer, server or data center via a wired network or a wireless network. The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media sets. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD) or a semiconductor medium. The semiconductor medium may be a solid-state hard disk.
[0150] 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. Professional and technical personnel 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.
[0151] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.
[0152] 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 only schematic, for example, the division of the units is only one, and there may be other division methods in actual implementation, 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.
[0153] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0154] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
[0155] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.
[0156] 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 protection scope of the present invention.
Claims
1. A personalized learning path planning method based on intelligent recommendation, characterized in that: The method comprises: According to the recorded online learning log, obtain the knowledge point set A that the target user has learned, and extract the knowledge point set B that the target user is interested in from the knowledge point set A; the knowledge point set A contains M learned knowledge points, and the knowledge point set B contains N interesting knowledge points, M≥N>0, M, N∈N + , N + is a positive integer; Obtain the target user's mastery level value for each knowledge point of interest, obtain N mastery level values, and filter out Q unmastered knowledge points of interest for the target user based on the mastery level values, Q>0, Q∈N + ; Acquire multiple non-target users similar to the target user, retrieve historical learning experience data of each non-target user, and screen out recommended learning video resources that are compatible with each unlearned knowledge point of interest of the target user based on the historical learning experience data to obtain a recommended course set; the historical learning experience data includes multiple learned knowledge points of the non-target user and selected watched learning video resources for each learned knowledge point; Inputting the recommended course set into a preconfigured genetic algorithm to plan an optimal learning path for the target user for the recommended course set, wherein the learning path represents a learning order of the recommended learning video resources; The method for obtaining the optimal learning path is as follows: e1: Initialize the population: randomly generate an original population, which contains X chromosomes, each chromosome represents a learning path, X>0, X∈N + ; e2: Fitness evaluation: Under each chromosome, obtain the target user's mastery gain value for each unmastered knowledge point of interest, input the mastery gain value of each unmastered knowledge point of interest into the pre-built fitness function, and calculate the fitness of each chromosome; The mastery degree gain value is calculated according to a predefined gain function, and its specific calculation formula is as follows: ; Where: represents the mastery gain value of the jth unmastered interest knowledge point, Indicates the degree of mastery after viewing. Indicates the initial mastery value, that is, the mastery value before watching; The calculation formula of the pre-constructed fitness function is as follows: ; Where: For fitness; is the total number of knowledge points of interest that have not been mastered; is the median function; The specific expression of the median function is as follows: ; Where: Represents the total number of mastery gain values of the jth unmastered interest knowledge point; e3: Selection: Use the roulette wheel method to select two chromosomes with high fitness in the original population as the father and mother; e4: Crossover: Perform crossover operation on the father and mother to generate new chromosomes; e5: Mutation: Perform mutation operation on the new chromosomes to obtain Y new chromosomes, combine the Y new chromosomes into a new population, replace the original population with the new population, and return to step e2; e6: Repeat steps e2 to e5 above until the fitness of the chromosome in the original population or the new population is greater than or equal to the preset fitness threshold, or the number of iterations is greater than or equal to the preset maximum number of iterations threshold, and output the learning path represented by the corresponding chromosome as the optimal learning path.
2. The personalized learning path planning method based on intelligent recommendation according to claim 1 is characterized in that: The step of extracting a set of knowledge points B that the target user is interested in includes: a1: Extract the target user's learning behavior information for the i-th learned knowledge point in the knowledge point set A, where the learning behavior information includes the target user's repeated learning times for the i-th learned knowledge point and the learning duration of each repeated learning, i>0, i∈N + ; a2: Input the learning behavior information of the i-th learned knowledge point into the pre-built interest measurement model to obtain the target user's interest coefficient for the i-th learned knowledge point; The mathematical expression of the pre-built interest measurement model is as follows: ; Where: is the interest coefficient of the i-th learned knowledge point; Indicates the number of times the target user repeats learning the i-th learned knowledge point; Indicates The learning time of the i-th learned knowledge point; is the time constant; is a natural constant; and are the sine and cosine functions respectively; is a logarithmic function; a3: Compare the interest coefficient of the i-th learned knowledge point with the preset interest coefficient threshold. If the interest coefficient is greater than or equal to the interest coefficient threshold, mark the i-th learned knowledge point as an interest knowledge point, set i=i+1, and return to step a1; if the interest coefficient is less than the interest coefficient threshold, set i=i+1 and directly return to step a1; a4: Repeat the above steps a1 to a3 until i=M, then end the loop, obtain N knowledge points of interest, and merge the N knowledge points of interest into a knowledge point set B.
3. The personalized learning path planning method based on intelligent recommendation according to claim 2 is characterized in that: The step of obtaining the target user's mastery level value for each knowledge point of interest includes: b1: Extract the jth interest knowledge point in the knowledge point set B, and retrieve the target user's assessment data for the jth interest knowledge point based on the historical test records, j>0, j∈N + ; The assessment data includes the assessment scores, assessment time and assessment error rate of the target user for the learned knowledge points in each special assessment; b2: Input the assessment data of the jth interest knowledge point into the pre-built mastery degree measurement model to obtain the mastery degree value of the target user for the jth interest knowledge point; set j=j+1, and return to step b1; The mathematical expression of the pre-built mastery degree measurement model is as follows: ; Where: is the mastery level value of the jth interest knowledge point; is the total number of special assessments for the target user on the jth knowledge point of interest; is the assessment score of the target user for the jth interest knowledge point in the vth special assessment; is the target user's error rate for the jth interest knowledge point in the vth special assessment; The time taken by the target user to assess the jth knowledge point of interest in the vth special assessment; is the time correction factor; b3: Repeat the above steps b1 to b2 until j=N, then end the loop and obtain the target user's mastery level value for each knowledge point of interest.
4. The personalized learning path planning method based on intelligent recommendation according to claim 3 is characterized in that: The step of screening out Q unlearned knowledge points of interest to the target user includes: c1: Normalize the mastery level of the gth interest knowledge point in the knowledge point set B, g>0, g∈N + ; c2: Compare the normalized mastery value of the g-th interest knowledge point with the preset mastery threshold. If the mastery value is greater than the mastery threshold, mark the g-th interest knowledge point as a mastered interest knowledge point, set g=g+1, and return to step c1; if the mastery value is less than or equal to the mastery threshold, mark the g-th interest knowledge point as an unmastered interest knowledge point, set g=g+1, and return to step c1; c3: Repeat the above steps c1 to c2 until g=M, then end the loop and obtain Q unmastered knowledge points of interest.
5. The personalized learning path planning method based on intelligent recommendation according to claim 4 is characterized in that: The step of screening out recommended learning video resources that are suitable for each unlearned knowledge point of interest of the target user includes: d1: extracting a knowledge point set C of non-target users from a plurality of learned knowledge points of non-target users; the knowledge point set C includes a plurality of knowledge points of interest that the non-target users have mastered; d2: Extract the acquired knowledge points in the knowledge point set C that are the same as the target user's Lth unacquired knowledge point, and mark them as overlapping knowledge points, L>0, L∈N + ; d3: Based on the learning video resources that non-target users choose to watch for each learned knowledge point, extract the learning video resources that are selected to watch for the overlapping knowledge points and mark them as candidate learning video resources; d4: Use the Apriori algorithm to calculate the support of each candidate learning video resource, sort them from large to small, and use the candidate learning video resource with the highest support as the recommended learning video resource that matches the target user's Lth unmastered interest point, and set L=L+1, and return to step d2; d5: Repeat steps d2 to d4 until L=Q, then end the loop and obtain recommended learning video resources that match each of the target user's unmastered knowledge points of interest. Merge the recommended learning video resources that match each of the target user's unmastered knowledge points of interest into a recommended course set.
6. A personalized learning path planning system based on intelligent recommendation, characterized in that: include: The first screening module is used to obtain the knowledge point set A that the target user has learned based on the recorded online learning log, and extract the knowledge point set B that the target user is interested in from the knowledge point set A; the knowledge point set A contains M learned knowledge points, and the knowledge point set B contains N interesting knowledge points, M≥N>0, M, N∈N + , N + is a positive integer; The second screening module is used to obtain the mastery degree value of each interest knowledge point of the target user, obtain N mastery degree values, and screen out Q unmastered interest knowledge points for the target user based on the mastery degree values, Q>0, Q∈N + ; The course recommendation module is used to obtain multiple non-target users similar to the target user, and retrieve the historical learning experience data of each non-target user, and screen out recommended learning video resources that are compatible with each unlearned knowledge point of interest of the target user according to the historical learning experience data, to obtain a recommended course set; the historical learning experience data includes multiple learned knowledge points of the non-target user and the learning video resources that have been selected and watched for each learned knowledge point; A path optimization module is used to input the recommended course set into a preconfigured genetic algorithm to plan the optimal learning path for the target user for the recommended course set, wherein the learning path represents the learning order of the recommended learning video resources; The method for obtaining the optimal learning path is as follows: e1: Initialize the population: randomly generate an original population, which contains X chromosomes, each chromosome represents a learning path, X>0, X∈N + ; e2: Fitness evaluation: Under each chromosome, obtain the target user's mastery gain value for each unmastered knowledge point of interest, input the mastery gain value of each unmastered knowledge point of interest into the pre-built fitness function, and calculate the fitness of each chromosome; The mastery degree gain value is calculated according to a predefined gain function, and its specific calculation formula is as follows: ; Where: represents the mastery gain value of the jth unmastered interest knowledge point, Indicates the degree of mastery after viewing. Indicates the initial mastery value, that is, the mastery value before watching; The calculation formula of the pre-constructed fitness function is as follows: ; Where: For fitness; is the total number of knowledge points of interest that have not been mastered; is the median function; The specific expression of the median function is as follows: ; Where: Represents the total number of mastery gain values of the jth unmastered interest knowledge point; e3: Selection: Use the roulette wheel method to select two chromosomes with high fitness in the original population as the father and mother; e4: Crossover: Perform crossover operation on the father and mother to generate new chromosomes; e5: Mutation: Perform mutation operation on the new chromosomes to obtain Y new chromosomes, combine the Y new chromosomes into a new population, replace the original population with the new population, and return to step e2; e6: Repeat steps e2 to e5 above until the fitness of the chromosome in the original population or the new population is greater than or equal to the preset fitness threshold, or the number of iterations is greater than or equal to the preset maximum number of iterations threshold, and output the learning path represented by the corresponding chromosome as the optimal learning path.
Citation Information
Patent Citations
Course arrangement method, electronic device, storage medium and program product
CN118333810A
