A method and system for recommending online music courses based on data processing

Abstract

This invention belongs to the field of data processing technology and provides an online music course recommendation method and system based on data processing. The method includes collecting first data from students, which includes at least their music interests and preferences, music level, instrument ownership, class schedule, and historical music assignment submission times. The first data is preprocessed to generate a multi-dimensional user profile, and a user-course interaction matrix is ​​generated based on the multi-dimensional user profile. The available music practice time for each student is determined based on the class schedule and historical music assignment submission times. Students are grouped into learning groups based on their music interests and preferences, music level, instrument ownership, and available music practice time. Based on the learning groups and the user-course interaction matrix, a course recommendation list is generated for each student, and the course recommendation list is pushed to the students.

Description

Technical Field

[0001] This invention belongs to the field of data processing technology, and specifically relates to an online music course recommendation method and system based on data processing. Background Technology

[0002] Online music courses, as part of internet education, have developed rapidly in recent years. Unlike traditional offline music teaching, online music courses are flexible and diverse, breaking through the limitations of time and space and providing convenient learning opportunities for students of different backgrounds and levels. These courses not only cover various music styles such as classical, jazz, pop, electronic, and folk, but also include learning on various instruments such as piano, guitar, violin, and vocals, as well as topics such as composition, arrangement, music theory, recording, and mixing. The curriculum covers everything from beginner to advanced professional levels, meeting the learning needs of different levels, including basic music theory learning, performance skill improvement, and music creation.

[0003] To enhance the user learning experience and help learners find more suitable courses, existing online music course recommendation systems typically employ data-driven recommendation methods. These systems generate user profiles by analyzing users' basic information, interests, preferences, and historical learning behaviors, and then recommend personalized courses to those users.

[0004] However, music courses require a lot of practice. Existing recommendation methods do not take into account the difficulties of online learning and offline practice, such as instrument availability and practice time. While peer-to-peer practice after online courses can partially solve the problem of practice difficulties, the existing course recommendation system does not take into account the differences in skill level and practice time among different students, making it difficult for students to form mutual support groups. Summary of the Invention

[0005] To address the problems in the existing technology, the present invention provides an online music course recommendation method based on data processing, the method comprising the following steps:

[0006] Collect students’ primary data, which includes at least their musical interests and preferences, musical level, musical instrument ownership, class schedule, and the submission time of their past music assignments.

[0007] The first data is preprocessed to generate a multi-dimensional user profile, and a user-course interaction matrix is ​​generated based on the multi-dimensional user profile.

[0008] The available time for students' music practice is determined based on the aforementioned course schedule and the submission time of historical music assignments.

[0009] Students will be grouped into study groups based on their musical interests, musical skills, musical instruments, and available time for musical practice.

[0010] Based on the learning group and the user-course interaction matrix, a course recommendation list is generated for each student, and the course recommendation list is pushed to the students.

[0011] In another aspect, the present invention provides an online music course recommendation system based on data processing, the system comprising the following modules:

[0012] The data collection module is used to collect students' primary data, which includes at least music interest preferences, music level, instrument ownership, class schedule, and historical music assignment submission time.

[0013] The processing module is used to preprocess the first data to generate a multi-dimensional user profile and generate a user-course interaction matrix based on the multi-dimensional user profile.

[0014] The generation module is used to determine the available time for students' music practice based on the timetable and the historical music assignment submission time.

[0015] The aggregation module is used to group students into learning groups based on their musical interests, musical level, musical instruments owned, and available time for musical practice.

[0016] The push module is used to generate a course recommendation list for each student based on the learning group and the user-course interaction matrix, and push the course recommendation list to the students.

[0017] The data-processing-based online music course recommendation method has the following beneficial effects:

[0018] By constructing multi-dimensional user profiles, taking into full account students' musical interests, skill levels, instrument ownership, and available practice time, a personalized course recommendation list is generated. This ensures that the recommended courses are highly matched to students' needs, improving the relevance and effectiveness of learning.

[0019] By grouping students with similar interests and abilities into learning groups, the potential for interaction and collaborative learning among students is enhanced. Group learning can promote knowledge sharing and skill improvement among students, increase overall learning efficiency, and enhance the enjoyment and sustainability of learning.

[0020] When recommending courses, we fully consider students' class schedules and the deadlines for submitting historical music assignments to ensure that recommended courses are reasonably scheduled within students' free time, thus guaranteeing the feasibility of the courses. This avoids time conflicts caused by unreasonable study arrangements and improves the operability of learning.

[0021] After the groups are formed, the system will dynamically adjust the structure of the learning groups based on the students' learning performance and changes in interests, so as to ensure that the students in each group have a high degree of similarity in terms of interests and learning abilities, and maximize the collaborative and interactive effects of group learning.

[0022] In summary, this recommendation method, by integrating user profiles, diverse recommendation algorithms, dynamic management of learning groups, and real-time feedback mechanisms, provides students with a more personalized, collaborative, and flexible music course learning solution, greatly improving students' learning outcomes and experience. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0025] The invention will now be described in preferred form with reference to the accompanying drawings and specific embodiments.

[0026] This embodiment solves the above problems through the following steps:

[0027] In one embodiment, reference Figure 1 This invention provides a method for recommending online music courses based on data processing. It involves a technical solution that generates a personalized course recommendation list for users by collecting and analyzing relevant user data, based on factors such as user interests, musical level, available practice time, and actual performance conditions.

[0028] The first data collected from students includes at least their musical interests and preferences, musical level, musical instrument ownership, class schedule, and the submission time of their past music assignments.

[0029] Collect the first data of the students, which includes at least the following:

[0030] Students' musical interests and preferences involve different musical styles and types, including but not limited to personal preferences for classical music, jazz, pop music, electronic music, folk music, rock music, blues music, country music, world music, etc., and further refine the information on interest preferences based on parameters such as specific music categories, composers, and performance styles.

[0031] The student's musical level includes their mastery of basic music theory, specifically the degree of mastery of fundamental music knowledge such as notes, rhythms, modes, chords, and intervals; their performance level on a specific instrument, such as the proficiency of their playing techniques, the types and difficulty levels of the instrumental techniques they have mastered, their sight-reading ability, their ability to read sheet music, and the complexity of the pieces they perform; and their ability in music composition, such as melody writing, harmony arrangement, rhythm composition, and musical arrangement.

[0032] The student's musical instrument ownership information includes whether the student owns a specific type of instrument, the type (e.g., piano, guitar, violin, drums, etc.), the quantity, model, and brand of the instrument, the condition of the instrument (e.g., newness, whether it is being used normally, maintenance status, etc.), and whether the student owns auxiliary equipment (e.g., pedals, speakers, amplifiers, etc.). The student's timetable information includes the student's current complete course schedule, not just music courses, but covering the specific arrangements of all enrolled courses, including weekly class times, class periods, course names, course types, course durations, and time arrangements with other learning activities. It further includes the student's daily free time. Combined with the course schedule, this allows for the determination of the student's free time and available time for music learning and practice throughout the day or week, thus providing a basis for developing a reasonable music learning plan and course recommendations.

[0033] The submission time of students' historical music assignments includes time data of various assignments submitted by students in the past music learning process. Specifically, it includes the submission date, the name of the course involved, the type of assignment (such as performance assignment, music theory assignment, composition assignment, etc.), the completion time of the assignment, the score of the assignment and the teacher's feedback information, as well as the time regularity of students' task completion and the timeliness of submission reflected in specific learning stages, so as to provide a basis for subsequent learning progress planning and recommendations.

[0034] The first data is preprocessed to generate a multi-dimensional user profile, and a user-course interaction matrix is ​​generated based on the multi-dimensional user profile.

[0035] In this step, the first data is preprocessed to generate a multi-dimensional user profile. The preprocessing includes data cleaning, missing value imputation, and data standardization and normalization, ultimately forming a high-quality and consistent user feature dataset. Based on this, a multi-dimensional user profile is further constructed, and a user-course interaction matrix is ​​generated according to the multi-dimensional user profile to support personalized recommendations.

[0036] First, the data on three dimensions—music interest preferences, music level, and instrument ownership—are preprocessed to better represent user characteristics when generating the user-course interaction matrix.

[0037] The user's music preferences are feature-encoded. The music style (such as classical, jazz, pop, etc.) is numerically represented using the TF-IDF (term frequency-inverse document frequency) method, resulting in a vector where each dimension represents the degree of user preference for a specific music style.

[0038] A standardized score is given to each user's musical level. This score is composed of multiple dimensions, including mastery of music theory, complexity of performance skills, and familiarity with the pieces, resulting in a comprehensive musical level score. The score is numerically converted between 0 and 1 and is used to reflect the user's current overall music level.

[0039] One-hot encoding is used to encode the types of musical instruments owned by the user (such as piano, guitar, violin, etc.), forming an instrument ownership vector. Each dimension represents whether the user owns a specific musical instrument, with a value of 0 or 1.

[0040] Constructing feature vectors for users and courses

[0041] Based on the above three features, the user's musical interests, musical level, and musical instrument ownership are combined to generate a user feature vector. ;

[0042] For each course, a feature vector is generated based on the course's content characteristics. in:

[0043] This is a vector representing the musical styles covered in the course, indicating their proportions.

[0044] The difficulty level of the course is rated, with a numerical value between 0 and 1, reflecting the course's requirements for musical level.

[0045] This is a vector representing the types of instruments required for the course, indicating the types of instruments needed to learn the course.

[0046] Content-based recommendation rating calculation

[0047] Content-based recommendations generate preliminary user preference ratings for courses by calculating the similarity between user and course feature vectors.

[0048] Calculate the user's music interest vector using cosine similarity. Course Music Style Vector

[0049] The similarity between them is expressed by the following formula:

[0050]

[0051] This value is used to measure a user's potential interest in this course style.

[0052] Musical level compatibility:

[0053] Calculate the user's music level rating Course difficulty rating Compatibility between them:

[0054]

[0055] The closer the fit is to 1, the better the user's music level matches the difficulty of the course.

[0056] Instrument compatibility:

[0057] User's musical instrument ownership status Instruments required for the course The matching degree between them is calculated using the vector dot product:

[0058] The dot product result indicates the degree of matching between the user and the required instruments for the course; the higher the value, the higher the degree of matching.

[0059] By combining the three similarity scores mentioned above, user X's content-based rating of course X is obtained. The weighted average method is used:

[0060]

[0061] in, The weight coefficients for different similarity scores satisfy the following conditions: It can be adjusted according to the actual scenario.

[0062] Furthermore, it captures the similarities and commonalities in learning habits among users.

[0063] The similarity between users is calculated based on their musical interests, musical skills, and musical instruments owned.

[0064] The Pearson correlation coefficient is used to measure the similarity between users:

[0065]

[0066] in, For user rating, the course is rated by user 𝑖. The average rating for user 𝑖 The average rating from users.

[0067] Based on historical ratings from similar users, target users For the course rating Make a prediction:

[0068]

[0069] This formula is used to predict a user's likely rating for a course they have not yet taken.

[0070] Content-based rating and rating prediction By integrating these elements, a final user-course interaction matrix is ​​generated.

[0071] The specific formula is as follows:

[0072]

[0073] in, and Let be the fusion coefficient, satisfying The fusion coefficient can be adjusted based on the model's performance to ensure that the recommendation results reflect both the user's content preferences and the experiences and behaviors of similar users.

[0074] Through the above steps and improvements, the generated user-course interaction matrix not only comprehensively considers the user's musical interests, skill level, and instrument ownership, but also ensures the accuracy, personalization, and dynamic adaptability of the recommendation results to changes in user interests.

[0075] The available time for students' music practice is determined based on the course schedule and the submission time of historical music assignments.

[0076] Based on the aforementioned timetable and historical music assignment submission times, the available time for students' music practice is determined. Specifically, the timetable data includes all of the student's current course schedules, the time slots, frequency, and duration of each course, while historical music assignment submission times include the dates and times of past assignment submissions. By combining this information, the student's study and free time is analyzed, and their available music practice time is scientifically and rationally allocated to provide a basis for personalized course recommendations.

[0077] To determine the available time for students' music practice, it is necessary to consider their class schedule and homework completion patterns, and use the following steps:

[0078] Course schedule data analysis

[0079] Time period division:

[0080] Divide the week into multiple time slots from morning to night, usually using 1 hour as a time slot (which can be adjusted to half an hour or 15 minutes as needed).

[0081] Construct a time matrix ,in This indicates whether a student has a class in the nth time slot on day n. If a class is scheduled, then... ,otherwise

[0082] Iterate through all time periods of each day and mark all... The time period is the student's free time period, and a list of all free time periods within a week is generated.

[0083] Analysis of historical assignment submission times

[0084] Assignment submission frequency statistics:

[0085] By statistically analyzing the submission times of historical music assignments, we can identify the practice time periods that students are more accustomed to.

[0086] Using time series analysis, we statistically analyzed the time points of assignment submissions to identify the time windows in a week when students frequently engaged in music learning and assignments, generating a list of priority time periods.

[0087] Determine practice time by combining free time and priority time.

[0088] The intersection of free time and prioritized practice time:

[0089] The optimal practice time for students is determined by finding the intersection of their free time on the timetable and the priority time slots for assignments. Specifically, this can be expressed as:

[0090]

[0091] in:

[0092] This indicates that the student's free time period is the nth time slot on day n.

[0093] This indicates the priority time for submitting the job during the X time slot on day X.

[0094] Based on the list of optimal practice times generated from the intersection results, the time slots within a week are sorted according to indicators such as the length of free time and the frequency of priority time, and finally a suitable time schedule for students to practice music is determined.

[0095] For example:

[0096] Suppose a student has the following class schedule and historical music assignment submission time information:

[0097] Course schedule information

[0098] School classes are held from 8:00 AM to 12:00 PM, Monday through Friday.

[0099] Additional subject tutoring sessions are available on Tuesdays and Thursdays from 2:00 PM to 4:00 PM.

[0100] All other time periods are considered free time.

[0101] Historical music assignment submission time

[0102] Over the past month, students have been accustomed to submitting their music assignments between 8:00 PM and 9:00 PM, at least three times a week.

[0103] The specific steps to generate available practice time are as follows:

[0104] Time matrix construction:

[0105] A week is divided into 24 hours per day, with each hour representing a separate time slot, for a total of 168 time slots.

[0106] Construct a time matrix, where the time period from 8:00 to 12:00 is Monday to Friday. Tuesdays and Thursdays from 2:00 PM to 4:00 PM other time periods

[0107] Idle time period marker:

[0108] From the time matrix , mark The time period is free time.

[0109] The available free time slots include 12:00 to 14:00 and 16:00 to 24:00 from Monday to Friday, and all day on weekends (0:00 to 24:00).

[0110] Priority practice time calculation:

[0111] Based on the submission times of historical music assignments, analysis revealed that 8:00 PM to 9:00 PM is the students' usual practice time. Priority time slots were generated and marked within a week.

[0112] Find the intersection:

[0113] Find the intersection of the idle time period and the priority time period:

[0114] Monday through Friday: The period from 8:00 PM to 9:00 PM is free time and also the historically preferred practice time, so it is marked as the best practice time.

[0115] Saturdays and Sundays are also free time, and given the habit of prioritizing practice, it is recommended to allocate some time for concentrated practice.

[0116] Determine the available time for practice:

[0117] The final generated music practice session has a available time of:

[0118] Monday to Friday, 8:00 PM to 9:00 PM.

[0119] Practice all day on Saturdays and Sundays, with a focus on practicing from 8:00 PM to 9:00 PM.

[0120] Final Practice Schedule

[0121] Based on the above analysis, the determined available timetable for student music practice is as follows:

[0122] Monday to Friday: 20:00 - 21:00.

[0123] Saturday and Sunday: Arrange the whole day flexibly, with a suggestion to focus on practice from 8:00 PM to 9:00 PM.

[0124] This approach integrates students' free time from classes with their historical learning habits, ensuring that practice time aligns with their daily rhythm while maintaining high feasibility and learning efficiency. This time allocation method helps recommendation systems provide more targeted music course recommendations, maximizing learning outcomes within the constraints of students' available practice time.

[0125] Students will be grouped into study groups based on their musical interests, musical skills, instrument ownership, and available time for musical practice.

[0126] To group students with similar interests, abilities, and schedules into study groups, a clustering algorithm is used to classify the students. The specific steps include:

[0127] Vector of each student's interests and preferences Music level score Instrument ownership vector Practice time vector Combined, they form a comprehensive feature vector. It is used for cluster analysis.

[0128] The similarity between the feature vectors of each pair of students is calculated. To accommodate the multidimensional nature of features, Euclidean distance or cosine similarity is used to measure the similarity between students. Euclidean distance is used to calculate the absolute differences of features across different dimensions, while cosine similarity is used to measure the consistency of students' interests and time management.

[0129] The K-means clustering algorithm was used to classify students. The specific steps are as follows:

[0130] Determine the number of clusters K: Based on the number of students and the distribution of different music styles and levels, select an appropriate number of clusters K so that the number of students in each group is moderate and easy to manage.

[0131] Randomly initialize centroids: Randomly select the feature vectors of K students as the initial cluster centers.

[0132] Iterative distance calculation: For each student, calculate the distance between them and each cluster center, and assign the student to the nearest cluster center.

[0133] Update centroids: For all students in each cluster, calculate the new centroids and update the centroids of each cluster.

[0134] Convergence condition judgment: Repeat the iteration until the cluster centers no longer change or the amount of change is lower than the set threshold, the clustering process ends, and K learning groups are finally formed.

[0135] Based on the clustering results, we further examine the feature matching degree of students within the group to ensure that students in each learning group have a high degree of consistency in terms of musical style interests, practice time arrangements, and instrument ownership, so as to ensure the synergistic effect within the learning group.

[0136] Based on students' music level scores Within a study group, some high-achieving students can be assigned as group leaders or assistant coaches to help other group members learn, thereby improving the overall learning efficiency and collaboration of the group.

[0137] This aggregation method ensures that members within each learning group have a high degree of similarity in terms of interests, skill level, musical instruments, and other aspects, in order to maximize collaborative learning and overall learning effectiveness.

[0138] Based on the learning group and the user-course interaction matrix, a course recommendation list is generated for each student, and the course recommendation list is pushed to the students.

[0139] Based on the learning groups and the user-course interaction matrix, a personalized course recommendation list is generated for each student. Specifically, firstly, information on students in the learning groups with similar interests, musical levels, and instrument ownership is compiled. Leveraging the collaborative learning potential among group members, the common needs of each group for music courses are analyzed. Simultaneously, combining the students' historical preferences and interest matching degrees for various courses recorded in the user-course interaction matrix, a rating prediction is generated for each student and all available courses.

[0140] In generating the course recommendation list, the system employs a combination of content-based recommendation algorithms. It analyzes students' interests in existing courses to recommend similar courses, and also leverages the learning activities of other group members, refining and optimizing the recommendations based on their choices and performance. The final course recommendation list for each student reflects not only their individual interests, level, and available study time, but also the learning consistency within the group, aiming to enhance student learning outcomes and interactive experiences through a group learning model.

[0141] The generated personalized course recommendation list will be sorted according to students' learning preferences, prioritizing courses that best suit their needs and have an appropriate level of difficulty at the top of the list. After the recommendations are completed, the system will push the recommendation list to each student in various ways, such as through the notification function of the online learning platform, email, or mobile application message reminders, so that students can easily view the recommended courses and promptly select suitable courses to start learning.

[0142] The entire recommendation process takes into account both individual and group characteristics of students, and the final course recommendation list pushed to students strikes a balance between personalization and collaboration, helping students better plan their music learning path and improve the continuity and effectiveness of their learning.

[0143] On the other hand, the present invention also provides an online music course recommendation system based on data processing, comprising:

[0144] The data collection module is used to collect students' primary data, which includes at least music interest preferences, music level, instrument ownership, class schedule, and historical music assignment submission time.

[0145] The processing module is used to preprocess the first data to generate a multi-dimensional user profile and generate a user-course interaction matrix based on the multi-dimensional user profile.

[0146] The generation module is used to determine the available time for students' music practice based on the timetable and the historical music assignment submission time.

[0147] The aggregation module is used to group students into learning groups based on their musical interests, musical level, musical instruments owned, and available time for musical practice.

[0148] The push module is used to generate a course recommendation list for each student based on the learning group and the user-course interaction matrix, and push the course recommendation list to the students.

[0149] For any module structures not specifically defined in this invention, the existing technical descriptions shall prevail. The prior art mentioned in the foregoing background and specific embodiments sections can be considered part of this invention and used to understand the meaning of certain technical features or parameters.

Claims

1. An online music course recommendation method based on data processing, characterized in that, The method comprises the following steps: Collecting first data of students, the first data at least comprising music interest preference, music level, instrument possession, timetable, historical music assignment submission time; Preprocessing the first data to generate a multi-dimensional user portrait, and generating a user-course interaction matrix according to the multi-dimensional user portrait; Determine the score of content by weighted summation based on the similarity between user and course feature vectors, the adaptation degree between user music level score and course difficulty score, and the matching degree between user instrument possession and required instruments of the course; Predict the score of the target user to the course based on the historical scores of similar users; Determine the user-course interaction matrix based on the score of content and the prediction result; Determine the available music practice time of students according to the timetable and historical music assignment submission time; Aggregate students into learning groups according to the music interest preference, music level, instrument possession, and available music practice time of students; The music style interest, practice time arrangement, and instrument possession of students in the learning group are consistent; Generate a course recommendation list for each student based on the learning group and the user-course interaction matrix, and push the course recommendation list to the students; Wherein, determining the available music practice time of students according to the timetable and historical music assignment submission time comprises: For each day, go through all time slots, mark all t ij = 0 as free time slots for the student, generate a list of all free times in a week; Assignment submission frequency statistics: Statistical analysis of historical music assignment submission time to identify the student's habit of practice time period; Use time series analysis to count the assignment submission time points, find out the time window of the student's frequent music learning and assignment within a week, and generate a preferred time period list; A student's optimal practice period A is determined by finding the intersection of the free periods of the schedule and the preferred time periods for homework ij , which is expressed as: A ij = T free (i,j)∩P preferred (i,j), Wherein: T free (i,j) indicates that the student has the jth time period on the ith day as a free time period; P preferred (i,j) represents the priority time of submitting a job in the jth time period of the ith day; According to the best practice time list generated according to the intersection result, sort the time period within a week according to the length of idle time and the frequency of preferred time indicators, and finally determine the timetable suitable for students to practice music; Generating a user-course interaction matrix according to the multi-dimensional user portrait comprises: Combine the user's music interest, music level, and instrument possession to generate a user feature vector U i = [I u , L u , E u ]; Wherein: I u a degree of preference of the user for the music genre; L u To the user's musical level; E u an instrument owned by the user; For each course, a feature vector C is generated based on the content characteristics of the course j = [I c , L c , E c ] Wherein: I c a music style vector for the course, indicating the music styles involved in the course and their proportions; L c To score the difficulty of the course, the value is between 0 and 1, reflecting the requirement of the course on the music level; E c a vector of types of instruments required for a course, indicating the types of instruments required to learn the course; computing a music interest vector I of the user u with a course music style vector I c : Calculating a musical level score L of the user u with the difficulty score L of the course: c between the two. S L (u,c) = 1 - |L u -L c | Calculating user instrument ownership E c The match degree between the required instruments for the course E u The match degree is calculated using vector dot product: S E (u,c) = E u ·E c ; Computing a content-based score of a course c by a user u Wherein, α1, α2, α3 are weight coefficients of different similarity scores, satisfying α1+α2+α3=1; Further, capture the similarity and commonness of learning habits between users; Compute similarity between users u i and users u k ​ where r ij is the rating of course j by user i, is the average rating of user i, is the average rating of user k; Based on the historical ratings of similar users, the rating of target user u i for course c j is predicted : generating a final user-course interaction matrix R = {r ij} ; Wherein, β1 and β2 are fusion coefficients, satisfying β1+β2=1.

2. The data processing based online music course recommendation method according to claim 1, wherein, The preprocessing comprises data cleaning, missing value filling, data standardization and normalization processing.

3. The data processing based online music course recommendation method of claim 1, wherein, When aggregating students into learning groups, the K-means clustering algorithm is used.

4. An online music course recommendation system based on data processing, the system implements the method of claim 1, characterized in that, The system comprises the following modules: A collection module for collecting first data of students, the first data at least comprising music interest preference, music level, instrument possession, timetable, historical music assignment submission time; A processing module for preprocessing the first data to generate a multi-dimensional user portrait, and generating a user-course interaction matrix according to the multi-dimensional user portrait; Determine the score of content by weighted summation based on the similarity between user and course feature vectors, the adaptation degree between user music level score and course difficulty score, and the matching degree between user instrument possession and required instruments of the course; predicting a score of a target user on a course based on historical scores of similar users; determining the user-course interaction matrix based on the content-based scores and the prediction results; generating a music practice available time of a student according to the course schedule and historical music assignment submission time; aggregating students into learning groups according to the music interest preference, music level, instrument possession and the music practice available time; the music style interest, practice time arrangement and instrument possession of students in the learning groups are consistent; pushing a course recommendation list to each student based on the learning groups and the user-course interaction matrix. determining the music practice available time of a student according to the course schedule and historical music assignment submission time comprises: For each day, all time periods are traversed, and all time periods with t ij = 0 are marked as free time periods for the student, and a list of all free times in a week is generated; assignment submission frequency statistics: counting historical music assignment submission time, analyzing the time regularity of students completing assignments, and identifying the practice time period of students; using time series analysis to count the assignment submission time points, finding the time window of students frequently learning music and assignments within a week, and generating a preferred time period list. The best practice time period A of the student is determined by finding the intersection of the free time of the course table and the priority time period of the homework ij , which is specifically represented as: A ij = T free (i,j)∩P preferred (i,j) wherein: T free (i,j) indicates that the student has a free time period on day i at time period j; P preferred (i,j) denotes the priority time of submitting a job at the jth time period of the ith day; the best practice time list generated according to the intersection result is sorted according to the length of idle time and the frequency of preferred time within a week, and finally a suitable time schedule for students to practice music is determined.

5. The online music lesson recommendation system based on data processing according to claim 4, wherein, The preprocessing includes data cleaning, missing value filling, data standardization and normalization processing.

6. The online music lesson recommendation system based on data processing according to claim 5, wherein, generating a user-course interaction matrix according to the multi-dimensional user portrait comprises: Combine the user's music interest, music level, and instrument possession to generate a user feature vector U i = [I u , L u , E u ]; wherein: I u a preference level of the user for the music genre; L u the user's musical level; E u an instrument owned by the user; For each course, a feature vector C is generated based on the content characteristics of the course j = [I c , L c , E c ] wherein: I c a music style vector for the course, representing the music styles involved in the course and their proportions; L c To score the difficulty of the course, the value is between 0 and 1, reflecting the requirement of the course on the music level; E c a vector of types of instruments required for a course, indicating the types of instruments required to learn the course; computing a music interest vector I of a user u with a course music style vector I c : Calculate the user's music level rating L u Course difficulty rating L c Compatibility between them: S L (u,c) = 1 - |L u -L c |, Calculating user instrument ownership E c The match degree between the required instruments for the course E u The match degree is calculated using vector dot product: S E (u,c) = E u ·E c Computing a content-based score of a course c by a user u wherein, α1, α2, α3 are weight coefficients of different similarity scores, satisfying α1+α2+α3=1; further, capturing the similarity between users and the commonness of learning habits; Compute similarity between users u i and users u k ​ where r ij is the rating of course j by user i, is the average rating of user i, is the average rating of user k; Based on the historical ratings of similar users, the rating of the target user u i for the course c j is predicted : generating a final user-course interaction matrix R = {r ij} wherein, β1 and β2 are fusion coefficients, satisfying β1+β2=1.

7. The online music lesson recommendation system based on data processing according to claim 6, wherein, When aggregating students into learning groups, the K-means clustering algorithm is used.

Images