Course recommendation method and system based on multidimensional data analysis

Abstract

The present invention provides a course recommendation method and system based on multidimensional data analysis, and relates to the technical field of course recommendation. By acquiring the target history data corresponding to the target user, a user-course matrix is ​​constructed according to the target history data to determine the candidate course information; the course fit is determined according to the candidate course information; the course selection data information is acquired and a hypergraph is constructed, and the course compatibility is determined through the hypergraph; the number of course selections and the average completion rate in the candidate course information are determined to generate a participation index, and popularity information is generated according to the participation index; the average score and user basic information corresponding to the candidate course are determined to generate the course difficulty; the comment area information is acquired in the candidate course information to determine the satisfaction; a comprehensive recommendation score is generated according to the course fit, course compatibility, popularity information, course difficulty and satisfaction. It provides students with more accurate, efficient and personalized course recommendations, which significantly improves students' learning experience and effect.

Description

Technical Field

[0001] The present application relates to the technical field of course recommendation, and in particular to a course recommendation method and system based on multidimensional data analysis. Background Art

[0002] Multidimensional data analysis is a powerful data processing and analysis method that can process data with multiple dimensions and reveal complex associations and patterns between data. In the field of education, the application of multidimensional data analysis technology can help to deeply understand students' learning behaviors and needs, thereby providing strong support for personalized education. Through multidimensional data analysis, the course recommendation system can comprehensively consider information from multiple dimensions such as students' historical learning data, course content, similarities between students, course popularity, difficulty, and user reviews, and recommend the most suitable courses for students.

[0003] Multidimensional data analysis is a method of processing and analyzing complex data, which can reveal hidden patterns and associations in the data. Multidimensional data analysis usually involves multiple dimensions, such as time, space, category, etc. Through the comprehensive analysis of multidimensional data, more comprehensive and profound insights can be obtained. Specifically for the course recommendation system, multidimensional data analysis can help analyze students' learning behaviors, course characteristics, similarities between students, etc., thereby providing strong support for personalized recommendations.

[0004] Therefore, how to use multidimensional data analysis to provide students with accurate course recommendations has become a technical problem that needs to be solved urgently. Summary of the invention

[0005] In order to provide students with accurate course recommendations by means of multidimensional data analysis, the present application provides a course recommendation method and system based on multidimensional data analysis.

[0006] In the first aspect, the present application provides a course recommendation method based on multidimensional data analysis using the following technical solutions:

[0007] A course recommendation method based on multidimensional data analysis, comprising:

[0008] Obtain target history data corresponding to the target user, and construct a user-course matrix based on the target history data to determine candidate course information;

[0009] Determining course suitability based on the candidate course information;

[0010] Obtaining course selection data information and constructing a hypergraph, and determining course compatibility through the hypergraph;

[0011] Determine the number of students enrolled and the average completion rate in the candidate course information to generate a participation index, and generate popularity information according to the participation index;

[0012] Determine the average score and user basic information corresponding to the candidate courses to generate the course difficulty;

[0013] Obtaining comment area information in the candidate course information to determine the satisfaction level;

[0014] A comprehensive recommendation score is generated according to the course fit, the course compatibility, the popularity information, the course difficulty and the satisfaction level.

[0015] Optionally, the step of obtaining target history data corresponding to the target user and constructing a user-course matrix according to the target history data to determine candidate course information includes:

[0016] Obtain target history data corresponding to the target user, and construct a user-course learning matrix based on the target history data;

[0017] Determine similar users in the user-course learning matrix using a user-based collaborative filtering recommendation algorithm;

[0018] Obtain the courses that the target user has learned and extract the subject words of the courses that have been learned through the LDA topic model to generate course features;

[0019] Candidate course information is determined according to the similar users and the course characteristics.

[0020] Optionally, the step of determining candidate course information according to the similar users and the course characteristics includes:

[0021] Determine, according to the similar users, a set of target courses that the similar users have studied but the target user has not studied as a user similarity calculation result;

[0022] Obtaining the learned course information of the target user, and determining the course similarity calculation result according to the learned course information and the course characteristics;

[0023] The candidate course information is generated according to the user similarity calculation result and the course similarity calculation result.

[0024] Optionally, the step of obtaining course selection data information and constructing a hypergraph, and determining course compatibility through the hypergraph includes:

[0025] Obtaining course selection data information, and determining node information and hyperedge information according to the course selection data information;

[0026] Constructing a hypergraph according to the node information and the hyperedge information;

[0027] converting the hypergraph into an incidence matrix;

[0028] identifying course combinations in the association matrix using a community detection algorithm;

[0029] Calculate the support, confidence and lift corresponding to the course combination;

[0030] The course compatibility is determined according to the support, the confidence, and the improvement.

[0031] Optionally, the step of determining the number of students enrolled and the average completion rate in the candidate course information to generate a participation index, and generating popularity information according to the participation index includes:

[0032] Obtain historical data of candidate courses, and determine the number of students enrolled in each course from the historical data and the average completion rate per course ;

[0033] Determine the number of students enrolled in each of the candidate courses according to the similar users ;

[0034] According to the number of students enrolled and the average completion rate Calculating the participation index

[0035] ;

[0036] According to the participation index Calculating popularity score

[0037] .

[0038] Optionally, the step of determining the average scores and user basic information corresponding to the candidate courses to generate the course difficulty includes:

[0039] Obtain historical performance information in the candidate course, and determine the course difficulty index DI based on the historical performance information

[0040]

[0041] in is the course average grade;

[0042] The learning ability index is calculated by combining the historical performance information using the weighted average method. LAI

[0043]

[0044] in is the grade of the i-th course, is the weight of the i-th course;

[0045] According to the course difficulty index DI and the learning ability index LAI Calculate the difficulty index for each course ADI

[0046]

[0047] in is the weight parameter.

[0048] Optionally, the step of obtaining comment area information in the candidate course information to determine the degree of satisfaction includes:

[0049] Acquire comment area information in the candidate course information, wherein the comment area information includes text information and non-text information;

[0050] Determining a word set in the text information, and determining a tag set in the word set according to a part-of-speech analysis strategy;

[0051] Calculating text sentiment scores in the tag set;

[0052] Determining non-text confidence information and sentiment classification ratio in the non-text information;

[0053] Determining a non-text sentiment score according to the non-text confidence information and the sentiment classification ratio;

[0054] A satisfaction level is determined based on the text sentiment score and the non-text sentiment score.

[0055] In a second aspect, the present application provides a course recommendation system based on multidimensional data analysis, the course recommendation system based on multidimensional data analysis comprising:

[0056] A candidate course information module is used to obtain target history data corresponding to a target user, and to construct a user-course matrix based on the target history data to determine candidate course information;

[0057] A course suitability module, used to determine course suitability according to the candidate course information;

[0058] A course compatibility module is used to obtain course selection data information and construct a hypergraph, and determine the course compatibility through the hypergraph;

[0059] A popularity information module, used for determining the number of students enrolled and the average completion rate in the candidate course information to generate a participation index, and generating popularity information according to the participation index;

[0060] A course difficulty module, used to determine the average scores and user basic information corresponding to the candidate courses to generate the course difficulty;

[0061] A satisfaction degree module, used for obtaining comment area information in the candidate course information to determine the satisfaction degree;

[0062] A comprehensive recommendation scoring module is used to generate a comprehensive recommendation score based on the course fit, the course compatibility, the popularity information, the course difficulty and the satisfaction level.

[0063] In a third aspect, the present application provides a computer device, comprising: a memory and a processor, wherein the processor executes the method described above when running computer instructions stored in the memory.

[0064] In a fourth aspect, the present application provides a computer-readable storage medium, comprising instructions, which, when executed on a computer, enable the computer to execute the method described above.

[0065] In summary, the present application includes the following beneficial technical effects:

[0066] This application obtains the target history data corresponding to the target user, builds a user-course matrix based on the target history data to determine the candidate course information; determines the course fit based on the candidate course information; obtains the course selection data information and constructs a hypergraph, and determines the course compatibility through the hypergraph; determines the number of course selections and the average completion rate in the candidate course information to generate the participation index, and generates popularity information based on the participation index; determines the average score and user basic information corresponding to the candidate course to generate the course difficulty; obtains the comment area information in the candidate course information to determine the satisfaction; and generates a comprehensive recommendation score based on the course fit, course compatibility, popularity information, course difficulty, and satisfaction. It provides students with more accurate, efficient, and personalized course recommendations, significantly improving their learning experience and results. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 It is a schematic diagram of the computer device structure of the hardware operating environment involved in the embodiment of the present application;

[0068] Figure 2 It is a flowchart of the first embodiment of the course recommendation method based on multidimensional data analysis of the present application;

[0069] Figure 3 It is a hypergraph schematic diagram of the first embodiment of the course recommendation method based on multidimensional data analysis of the present application;

[0070] Figure 4 It is a structural block diagram of the first embodiment of the course recommendation system based on multidimensional data analysis of the present application. DETAILED DESCRIPTION

[0071] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below through the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0072] Reference Figure 1 , Figure 1 A schematic diagram of the computer device structure of the hardware operating environment involved in the embodiment of the present application.

[0073] like Figure 1 As shown, the computer device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a wireless fidelity (Wireless-Fidelity, Wi-Fi) interface). The memory 1005 may be a high-speed random access memory (Random Access Memory, RAM), or a stable non-volatile memory (Non-Volatile Memory, NVM), such as a disk storage. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0074] Those skilled in the art will understand that Figure 1 The structure shown in the figure does not constitute a limitation on the computer device, and may include more or less components than shown in the figure, or combine some components, or arrange the components differently.

[0075] like Figure 1 As shown, the memory 1005 as a storage medium may include an operating system, a network communication module, a user interface module, and a course recommendation program based on multi-dimensional data analysis.

[0076] exist Figure 1 In the computer device shown, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and the memory 1005 in the present application can be set in the computer device, and the computer device calls the course recommendation program based on multidimensional data analysis stored in the memory 1005 through the processor 1001, and executes the course recommendation method based on multidimensional data analysis provided in the embodiment of the present application.

[0077] The present application embodiment provides a course recommendation method based on multidimensional data analysis. Figure 2 , Figure 2 This is a flowchart of the first embodiment of the course recommendation method based on multidimensional data analysis of this application.

[0078] In this embodiment, the course recommendation method based on multidimensional data analysis includes the following steps:

[0079] Step S10: Acquire the target history data corresponding to the target user, and construct a user-course matrix based on the target history data to determine candidate course information.

[0080] It should be noted that the steps of obtaining the target history data corresponding to the target user and constructing a user-course matrix based on the target history data to determine the candidate course information include: obtaining the target history data corresponding to the target user and constructing a user-course learning matrix based on the target history data; determining similar users in the user-course learning matrix using a user-based collaborative filtering recommendation algorithm; obtaining the courses that the target user has learned and extracting the subject words of the learned courses through the LDA topic model to generate course features; and determining the candidate course information based on similar users and course features.

[0081] It can be understood that the step of determining candidate course information based on similar users and course characteristics includes: determining a set of target courses that similar users have studied but the target user has not studied as a user similarity calculation result based on similar users; obtaining the target user's learned course information, and determining the course similarity calculation result based on the learned course information and the course characteristics; generating candidate course information based on the user similarity calculation result and the course similarity calculation result.

[0082] In the specific implementation, based on the historical behavior data of the target user u, a user-course learning matrix is ​​constructed (the elements in the matrix indicate whether the user has selected a course, 1 indicates selection, and 0 indicates non-selection). The user-based collaborative filtering recommendation algorithm is used to calculate the similar users v of the target user u. The calculation formula for similar users is:

[0083]

[0084] in, and Represent the course sets selected by user u and user v respectively.

[0085] Extract the course introduction of the target user's already learned course i, use the LDA topic model to extract the course's subject words, and vectorize them. Use the extracted course features to calculate the similarity between the already learned course i and course j. The calculation formula is:

[0086]

[0087] in, and denote the kth feature of course i and course j respectively.

[0088] Based on the similar users calculated above, select the K users K(u) most similar to the target user, and extract the courses that the similar users have learned but the target user has not learned. The formula is:

[0089]

[0090] in, represents the set of courses selected by user v, Represents the set of courses selected by the target user u.

[0091] Calculate the similar courses between the target user and the courses he has taken. For each course i that the target user has taken, calculate the most similar course j, and select the similar courses that have not been taken as candidate courses. The formula is:

[0092]

[0093] in, A collection of courses for similar users.

[0094] It should be noted that similarity between students is also one of the important dimensions of multidimensional data analysis. By analyzing students' background information, learning behaviors, and academic performance, the system can identify groups of students with similar learning needs and interests. In this way, the system can draw on the learning paths and experiences of similar students to provide references for individual students. For example, if a student is very similar to a group of high-performing students in learning behavior, the system can refer to the courses selected by these high-performing students and recommend similar courses to the student to improve his learning effect.

[0095] It is understandable that students' historical learning data is one of the important bases for the course recommendation system. This data includes the courses that students have previously taken, their academic performance, study time, completed tasks and homework, etc. By analyzing this data, the system can understand students' learning styles and habits, and predict their interest and adaptability to certain courses. For example, if a student has shown a strong interest in programming courses in the past and has excellent grades, the system can give priority to recommending more advanced programming courses or related project courses to meet the student's learning needs and interests.

[0096] In specific implementation, collaborative filtering algorithm is a common recommendation system algorithm, which is widely used in e-commerce, social media, content recommendation and other fields. Its core idea is to mine the similarity between users or the similarity between items based on the historical behavior data of users and make recommendations. Collaborative filtering algorithms are mainly divided into two categories: user-based collaborative filtering and item-based collaborative filtering. User-based collaborative filtering is a classic recommendation algorithm that recommends courses that similar users like by calculating the similarity between users. This method assumes that users with similar interests will choose similar courses. Therefore, by finding users similar to the target user and recommending courses that these users like, the accuracy of the recommendation can be improved. In practical applications, user-based collaborative filtering usually uses a user-course rating matrix. Each row of the matrix represents a user, each column represents a course, and the elements in the matrix represent the user's rating of the course. In order to calculate the similarity between users, a variety of similarity measurement methods can be used, such as cosine similarity, Pearson correlation coefficient, etc. After calculating the similarity, the target user's rating of the unrated course can be predicted based on the ratings of similar users. For example, for the target user A, find the K users (called neighbor users) most similar to A, and then calculate the predicted score of the course for A based on the ratings of these neighbor users on a certain course. Item-based collaborative filtering is another important method of collaborative filtering. Unlike user-based collaborative filtering, item-based collaborative filtering focuses on the similarity between courses and recommends courses similar to the courses selected by the user. This method assumes that users will be interested in courses similar to their favorite courses. Therefore, by calculating the similarity between courses, similar courses can be recommended to users. In item-based collaborative filtering, it is also necessary to construct a user-course rating matrix. The difference is that this embodiment focuses on the similarity between the columns of the matrix, rather than the similarity between the rows. The similarity between courses can be calculated by a variety of methods, such as cosine similarity, Pearson correlation coefficient, etc. Once the similarity between courses is calculated, the user's rating of unselected courses can be predicted based on the ratings of the user's selected courses. For example, for the target user A and course i, A's rating of i can be predicted based on A's ratings of other courses and the similarity of these courses with i.

[0097] Step S20: Determine course suitability based on candidate course information.

[0098] In the specific implementation, the recommendation results based on user similarity and course similarity are combined to obtain the final candidate course list. Using the weighted method, the two recommendation results are combined and the K sequences with the highest scores are selected to form the final candidate courses. The specific formula is as follows:

[0099]

[0100] in, is a weight parameter whose value is within [0,1].

[0101] Step S30: Obtain course selection data information and construct a hypergraph, and determine course compatibility through the hypergraph.

[0102] It should be noted that the steps of obtaining course selection data information and constructing a hypergraph, and determining course compatibility through the hypergraph include: obtaining course selection data information, determining node information and hyperedge information based on the course selection data information; constructing a hypergraph based on the node information and hyperedge information; converting the hypergraph into an association matrix; determining course combinations in the association matrix using a community detection algorithm; calculating support, confidence, and lift corresponding to the course combinations; and determining course compatibility based on the support, confidence, and lift.

[0103] In a specific implementation, the number of students enrolled and the average completion rate in the candidate course information are determined to generate a participation index, and the steps of generating popularity information according to the participation index include: obtaining historical data of the candidate courses, determining the number of students enrolled in each course in the historical data and the average completion rate per course ; Determine the number of students enrolled in each course in the candidate courses based on similar users ; According to the number of students enrolled and average completion rate Calculating the participation index

[0104] ;

[0105] According to the participation index Calculating popularity score

[0106] .

[0107] In the specific implementation, nodes are defined, and each course is defined as a node. For example, course A, course B, course C, etc. are defined as different nodes.

[0108] Define hyperedges based on two criteria: subject similarity and user common selection. Subject similarity: Calculate the similarity score of course content. If the content of two courses is highly similar, add a hyperedge connecting the two courses to the hypergraph. Common selection: Count the frequency of students taking two or more courses at the same time. If the frequency is high, add a hyperedge connecting these courses to the hypergraph. Figure 3 As shown, , , , It is a hyperedge. Different area colors represent different nodes included in the hyperedge. For the node.

[0109] Convert the hypergraph into an incidence matrix, where each row represents a node (course) and each column represents a hyperedge. The elements of the incidence matrix are 0 or 1, indicating whether the node exists in the corresponding hyperedge. Definition , means if i belongs to hyperedge j, otherwise . As shown in the following table:

[0110]

[0111] Select a course combination and use the community detection algorithm to first optimize the modularity. Calculate the modularity of the association matrix and iteratively adjust the node allocation to maximize the modularity Q. The calculation formula is as follows:

[0112]

[0113] Where B is the weighted adjacency matrix calculated from the incidence matrix H, and are the degrees of nodes i and j respectively, m is the total number of edges in the graph, is an indicator function of whether nodes i and j belong to the same community.

[0114] Secondly, during label propagation, each node's label is initialized to its own. During the iteration process, each node is updated to the label with the highest frequency among its neighboring nodes until the label no longer changes. Finally, based on the algorithm results, multiple communities (course combinations) are identified, each of which contains a group of closely related courses.

[0115] Based on the course combinations calculated by the above algorithm, calculate support, confidence, and lift. Evaluate the compatibility of each course group. Course groups with high support, high confidence, and lift greater than 1 indicate high compatibility and are recorded as candidate courses. The calculation formula is as follows:

[0116] Support:

[0117]

[0118] Confidence:

[0119]

[0120] Lift:

[0121]

[0122] Among them, A and B are courses, T is the course selection record of all students, and each element Represents a student's course selection set. For example, if a student takes course A and course B, then the course selection record of this student Contains A and B. It is a counter that indicates the number of students who take both Course A and Course B. It counts the number of records that contain both Course A and Course B in all student course selection records T.

[0123] It should be noted that a hypergraph is an extended form of a graph, which is used to represent multivariate relationships. In traditional graph models, edges can only connect two nodes, while in hypergraph models, hyperedges can connect multiple nodes, thereby better representing complex multivariate relationships. Hypergraph models have significant advantages in processing multidimensional data analysis, especially in course recommendation systems, where they can be used to represent complex associations between courses and students' course selection behaviors. In course recommendation systems, nodes usually represent courses, while hyperedges represent the associations between courses. In course recommendation, the hypergraph model models the complex relationships between students and courses by constructing a hypergraph structure that includes students, courses, and their related features. In this hypergraph, nodes represent students and courses, and hyperedges represent courses selected or evaluated by students and the association features between courses (such as course content similarity, course difficulty, etc.). By analyzing and calculating the hypergraph, students' interest preferences and course characteristics can be captured more accurately, thereby achieving more accurate course recommendations and improving the effectiveness of the recommendation system and students' learning experience.

[0124] Step S40: Determine the number of students enrolled and the average completion rate in the candidate course information to generate a participation index, and generate popularity information according to the participation index.

[0125] In the specific implementation, obtain the historical data of candidate courses and the number of students enrolled in each course and the average completion rate per course ;

[0126] Select the top 10 similar users and count the number of students enrolled in the candidate courses .

[0127] According to the number of students enrolled in each course and the average completion rate per course Calculate the course participation index for existing courses , the calculation formula is as follows:

[0128]

[0129] Calculate the final course popularity score , the calculation formula is as follows:

[0130]

[0131] For the popularity of new courses (cold start), this embodiment adopts similar course analysis, calculates the similarity of course content and course introduction, selects the five courses with the highest similarity, calculates the popularity scores of these five courses, and takes the average as the popularity score of the course.

[0132] Step S50: Determine the average scores and user basic information corresponding to the candidate courses to generate the course difficulty.

[0133] It should be noted that the step of determining the average score and user basic information corresponding to the candidate courses to generate the course difficulty includes: obtaining historical score information in the candidate courses, and determining the course difficulty index DI according to the historical score information

[0134]

[0135] in The average grade of the course; the weighted average method is used to combine historical grade information to calculate the learning ability index LAI

[0136]

[0137] in is the grade of the i-th course, is the weight of the i-th course; according to the course difficulty index DI and learning ability index LAI Calculate the Adaptive Difficulty Index (ADI) for each course

[0138]

[0139] in is the weight parameter.

[0140] In specific implementation, the popularity and difficulty of a course are also important factors in course recommendation. The popularity of a course can reflect the quality and popularity of the course, while the difficulty of a course can help students plan their learning path reasonably and avoid taking courses that are too difficult or too easy. For example, the system can recommend some popular and well-reviewed courses to help students find high-quality learning resources. At the same time, the system can also recommend courses of appropriate difficulty based on the students' learning ability and level, ensuring that students can challenge themselves during the learning process without feeling excessive pressure.

[0141] Step S60: Obtain the comment area information in the candidate course information to determine the satisfaction level.

[0142] It can be understood that the step of obtaining comment area information in the candidate course information to determine the degree of satisfaction includes: obtaining comment area information in the candidate course information, the comment area information including text information and non-text information; determining a word set in the text information, and determining a tag set in the word set according to a part-of-speech analysis strategy; calculating a text sentiment score in the tag set; determining non-text confidence information and sentiment classification ratio in the non-text information; determining a non-text sentiment score based on the non-text confidence information and the sentiment classification ratio; and determining the degree of satisfaction based on the text sentiment score and the non-text sentiment score.

[0143] It should be noted that in this embodiment, text information includes text comment information, etc., and non-text information includes but is not limited to emoji expressions, which is not limited in this embodiment.

[0144] In the specific implementation, the text and emoji expressions in the candidate course comment area are extracted separately. The extraction format of each comment is: "text" + "emoji expression", and they are processed separately.

[0145] For text reviews, we first perform Chinese word segmentation to divide the sentence into multiple words. For example, for the comment "the course is very good", the word segmentation result may be "course / very / good". We perform word part disambiguation on the word segmentation results to determine the specific part of speech of each word, and identify the combination of nouns and adjectives in the comments, and use adjectives as pseudo labels for the courses. We analyze the dependency relationship between words to ensure that there is a modification relationship between adjectives and nouns, such as "good" modifies "course".

[0146] The extracted evaluation words are scored, and the scoring is expanded using dictionaries such as "Harbin Institute of Technology Synonyms Dictionary". Combined with manual scoring, positive words are given positive scores and negative words are given negative scores, such as "good" is scored as 6 and "bad" is scored as -6. User comments are processed to obtain the satisfaction score of a single comment , the calculation formula is as follows:

[0147]

[0148] in, is the score of the kth evaluation word, and m is the total number of evaluation words.

[0149] For emoji expressions, the confidence and sentiment classification ratio of each expression package are extracted according to the "Emoji Sentiment Analysis List". Each expression package can be expressed in the following form:

[0150]

[0151] Among them, the negative ratio is assigned a negative value. The neutral ratio is a smoothing factor, and the sign is determined according to the part of speech of the comment text. If the comment is a positive comment, it is assigned a positive value, otherwise it is assigned a negative value. The positive ratio is assigned a positive value.

[0152] Use the "Emoji Sentiment Analysis List" to calculate its sentiment value. According to the table, the sentiment value is a vector and can be expressed as:

[0153]

[0154] For example, the emoticon "Hehe" (character encoding: U+1F600):

[0155] Confidence:

[0156] Negative percentage: 9.38%

[0157] Neutral percentage: 33.64%

[0158] Active percentage: 56.98%

[0159] but:

[0160] For each emoji in the comment, calculate its sentiment value and take the average:

[0161]

[0162] in, is the sentiment value of the lth emoticon package, and p is the total number of emoticons.

[0163] Calculate the comprehensive satisfaction score of a single comment, combine the text sentiment score and the emoticon sentiment score to get the comprehensive satisfaction score of a single comment The calculation formula is as follows:

[0164]

[0165] in, is the weight coefficient.

[0166] Calculate the overall satisfaction score for the course, the average of all reviews' satisfaction scores The calculation formula is:

[0167]

[0168] in, is the satisfaction score of the jth review, and n is the total number of reviews for the course.

[0169] Step S70: Generate a comprehensive recommendation score based on course fit, course compatibility, popularity information, course difficulty, and satisfaction.

[0170] In the specific implementation, the weight parameters of each indicator are determined, and the weights of user fit, course compatibility, course popularity, course difficulty adaptation and user review satisfaction are set based on actual data and user needs.

[0171] Each candidate course is scored for each indicator. User fit, course compatibility, course popularity, course difficulty adaptation, and user review satisfaction are calculated based on the corresponding calculation results in the previous steps.

[0172] The scores of the above five indicators are weighted and summed according to the weight parameters to calculate the comprehensive score of each candidate course. The specific formula is as follows:

[0173]

[0174] in, represents the comprehensive score of the candidate courses, are the weights of user fit, course compatibility, course popularity, course difficulty adaptation, and user review satisfaction, respectively. represents the user fit score, represents the course compatibility score, represents the course popularity score, Indicates the course difficulty suitability score, Represents the user review satisfaction rating.

[0175] The candidate courses are sorted according to the comprehensive scores, and the top K courses with the highest comprehensive scores are selected as the course list recommended to the user.

[0176] Dynamically adjust weight parameters, and regularly update weight parameters and recommendation models based on user feedback and learning effects to continuously improve the accuracy of the recommendation system and user satisfaction.

[0177] It should be noted that this embodiment automatically recommends the most suitable courses for students by analyzing students' learning behaviors and preferences, thereby improving students' learning effects and satisfaction. In particular, with the support of multidimensional data analysis technology, it is possible to comprehensively consider multiple factors and provide students with more accurate and personalized course recommendations. Through multidimensional data analysis, the course recommendation system can comprehensively consider information from multiple dimensions such as students' historical learning data, course content, similarities between students, course popularity, difficulty, and user evaluations, and recommend the most suitable courses for students.

[0178] It can be understood that the main advantages of this embodiment include:

[0179] Comprehensive application of multi-dimensional evaluation indicators: This embodiment uses five key evaluation indicators: user fit, course compatibility, course popularity, course difficulty and user comments. These indicators are optimized comprehensively to improve the accuracy of course recommendations and user satisfaction. Multi-dimensional evaluation can comprehensively analyze all aspects of courses and users to ensure that recommended courses are more in line with user needs and interests. User fit is calculated by analyzing the user's historical behavior and the choices of similar users to ensure that the recommended courses are highly matched with the user's interests. Course compatibility analyzes the similarity between courses through a hypergraph model to help identify potentially compatible course combinations. Course popularity is evaluated by the number of course selections and completion rates to ensure that the recommended courses are popular and of high quality. Course difficulty is calculated by average grades and the user's basic level to ensure that the recommended courses are suitable for the user's learning ability. User comments quantify satisfaction through sentiment analysis and ratings, and use feedback from other users to help new users make more informed choices.

[0180] Personalized recommendations: This embodiment uses item-based and user-based collaborative filtering algorithms to calculate the fit between users and candidate courses. By analyzing the user's historical behavior data and the choices of similar users, the system can accurately recommend courses that meet the user's interests and needs. This personalized recommendation can greatly improve the user's learning experience and participation. For example, by analyzing the courses previously selected and the grades achieved by the user, the system can identify the user's learning interests and tendencies, and recommend similar or related courses. The user-based collaborative filtering algorithm can find users similar to the target user and recommend courses that these users like. The item-based collaborative filtering algorithm can find courses similar to the user's selected courses and recommend them. Combining these two methods, the system can provide more accurate and personalized course recommendations.

[0181] Course compatibility analysis: This embodiment uses a hypergraph model to analyze the compatibility between courses. By analyzing the user co-selection behavior and prerequisite relationships between courses, the system can identify potentially compatible course combinations and support interdisciplinary learning. This analysis helps students choose suitable course combinations, improve learning outcomes and the rationality of the overall knowledge structure. The hypergraph model can not only identify direct similarities between courses, but also identify indirect similarities and potential course combinations through users' course selection behavior and prerequisite relationships. This compatibility analysis can help students conduct interdisciplinary learning, choose course combinations that complement and enhance each other, and improve the breadth and depth of learning.

[0182] Comprehensive evaluation of popularity and difficulty: This embodiment evaluates the popularity of the course by counting the number of course enrollments and completion rates, and calculates the course difficulty by the average course score and the user's foundation and level. Popularity evaluation ensures that the recommended courses are widely recognized, while difficulty evaluation ensures that the recommended courses are adapted to the students' learning level, avoiding the recommendation of courses that are too difficult or too low. Courses with high popularity are usually of high quality and are popular with most students. Recommending these courses can improve the user's learning satisfaction. Course difficulty evaluation can ensure that the recommended courses are adapted to the user's learning ability and level, avoid recommending courses that are too difficult or too low, and ensure that users can successfully complete the courses and achieve good learning results.

[0183] User review analysis: This embodiment quantifies user satisfaction with the course through sentiment analysis and scoring. Adjectives and sentiment words in user reviews are extracted and assigned values ​​to form a satisfaction score for the course. With real feedback from other users, new users can make more informed choices, thereby improving their learning experience. For example, by analyzing positive and negative sentiment words in user reviews, the system is able to calculate the satisfaction score for each course. Courses with more positive reviews have higher scores, and vice versa. Users can use these scores to understand the actual situation of the course and the experience of other users, and make more informed course selection decisions. This analysis based on user reviews not only provides an objective evaluation, but also enhances the credibility and practicality of the recommendations.

[0184] Dynamic adjustment and feedback mechanism: This embodiment will regularly update the weight parameters and recommendation model based on user feedback and learning effects. This dynamic adjustment mechanism ensures that the system can continuously improve the accuracy of recommendations and user satisfaction, and adapt to the changing needs of users. By collecting user feedback and learning data, the system can understand the effect of recommendations and user satisfaction, and timely adjust the recommendation algorithm and weight parameters to optimize the recommendation model. For example, if the recommendation effect of certain courses is not good, and the user's academic performance and satisfaction are low, the system will reduce the weight of these courses and reduce their recommendation probability. On the contrary, for those courses with good user feedback and good learning effects, the system will increase their weights and increase their recommendation probability. Through this dynamic adjustment and optimization, the system can continuously improve the effect of recommendations and user satisfaction, and maintain the efficiency and accuracy of the recommendation system.

[0185] Improve learning outcomes: This embodiment effectively improves students' learning outcomes through accurate recommendations. Students no longer need to get lost in a large number of courses, and can find courses that suit them more efficiently, thereby saving time and energy and improving learning efficiency and effectiveness. Accurate recommendations can recommend the most suitable courses based on students' interests and learning needs, helping students concentrate on learning and improve learning outcomes. For example, for students studying programming, the system can recommend high-quality courses related to programming to help students systematically learn programming knowledge and skills and achieve better learning outcomes. Through this accurate recommendation, students can acquire more knowledge and skills in a short period of time and improve their learning efficiency and effectiveness.

[0186] Support for multiple learning paths: This embodiment can identify highly compatible course combinations to support students' diverse learning needs. Students can choose different learning paths based on their interests and learning goals to obtain a more personalized and flexible learning experience. By analyzing the compatibility between courses and students' learning behaviors, multiple learning paths can be recommended to meet students' different learning needs and goals. For example, for students who want to have a comprehensive understanding of computer science, the system can recommend multiple courses related to computer science to help students learn comprehensively and systematically. For students who want to focus on a certain field, the system can recommend in-depth courses in that field to help students study and research in depth. By supporting multiple learning paths, the system can meet students' different learning needs and goals and provide a more flexible and personalized learning experience.

[0187] This embodiment obtains the target history data corresponding to the target user, builds a user-course matrix based on the target history data to determine the candidate course information; determines the course fit based on the candidate course information; obtains the course selection data information and builds a hypergraph, and determines the course compatibility through the hypergraph; determines the number of course selections and the average completion rate in the candidate course information to generate the participation index, and generates popularity information based on the participation index; determines the average score and user basic information corresponding to the candidate course to generate the course difficulty; obtains the comment area information in the candidate course information to determine the satisfaction; and generates a comprehensive recommendation score based on the course fit, course compatibility, popularity information, course difficulty, and satisfaction. It provides students with more accurate, efficient, and personalized course recommendations, significantly improving their learning experience and results.

[0188] In addition, an embodiment of the present application also proposes a computer-readable storage medium, on which is stored a program for course recommendation based on multidimensional data analysis. When the program for course recommendation based on multidimensional data analysis is executed by a processor, the steps of the method for course recommendation based on multidimensional data analysis as described above are implemented.

[0189] Reference Figure 4 , Figure 4This is a structural block diagram of the first embodiment of the course recommendation system based on multidimensional data analysis of this application.

[0190] like Figure 4 As shown, the course recommendation system based on multidimensional data analysis proposed in the embodiment of the present application includes:

[0191] The candidate course information module 10 is used to obtain the target history data corresponding to the target user, and to construct a user-course matrix according to the target history data to determine the candidate course information;

[0192] A course suitability module 20, for determining course suitability based on candidate course information;

[0193] The course compatibility module 30 is used to obtain course selection data information and construct a hypergraph, and determine the course compatibility through the hypergraph;

[0194] A popularity information module 40 is used to determine the number of course enrollments and the average completion rate in the candidate course information to generate a participation index, and to generate popularity information according to the participation index;

[0195] The course difficulty module 50 is used to determine the average scores and user basic information corresponding to the candidate courses to generate the course difficulty;

[0196] A satisfaction degree module 60 is used to obtain comment area information in the candidate course information to determine the satisfaction degree;

[0197] The comprehensive recommendation rating module 70 is used to generate a comprehensive recommendation rating based on course fit, course compatibility, popularity information, course difficulty and satisfaction.

[0198] It should be understood that the above is only an example and does not constitute any limitation on the technical solution of the present application. In specific applications, technicians in this field can make settings as needed, and the present application does not impose any limitation on this.

[0199] This embodiment obtains the target history data corresponding to the target user, builds a user-course matrix based on the target history data to determine the candidate course information; determines the course fit based on the candidate course information; obtains the course selection data information and builds a hypergraph, and determines the course compatibility through the hypergraph; determines the number of course selections and the average completion rate in the candidate course information to generate the participation index, and generates popularity information based on the participation index; determines the average score and user basic information corresponding to the candidate course to generate the course difficulty; obtains the comment area information in the candidate course information to determine the satisfaction; and generates a comprehensive recommendation score based on the course fit, course compatibility, popularity information, course difficulty, and satisfaction. It provides students with more accurate, efficient, and personalized course recommendations, significantly improving their learning experience and results.

[0200] In one embodiment, the candidate course information module 10 is further used to obtain target history data corresponding to the target user and construct a user-course learning matrix according to the target history data;

[0201] A user-based collaborative filtering recommendation algorithm is used to identify similar users in the user-course learning matrix; the courses that the target user has learned are obtained and the subject words of the learned courses are extracted through the LDA topic model to generate course features; candidate course information is determined based on similar users and course features.

[0202] In one embodiment, the candidate course information module 10 is also used to determine a set of target courses that similar users have studied but the target user has not studied as a user similarity calculation result based on similar users; obtain the target user's learned course information, and determine the course similarity calculation result based on the learned course information combined with course characteristics; generate candidate course information based on the user similarity calculation result and the course similarity calculation result.

[0203] In one embodiment, the course compatibility module 30 is also used to obtain course selection data information, determine node information and hyperedge information based on the course selection data information; construct a hypergraph based on the node information and hyperedge information; convert the hypergraph into an association matrix; use a community detection algorithm to determine course combinations in the association matrix; calculate the support, confidence, and lift corresponding to the course combination; and determine the course compatibility based on the support, confidence, and lift.

[0204] In one embodiment, the course compatibility module 30 is further used to obtain historical data of candidate courses and determine the number of students enrolled in each course from the historical data. and the average completion rate per course ; Determine the number of students enrolled in each course in the candidate courses based on similar users ; According to the number of students enrolled and average completion rate Calculating the participation index

[0205] ; Based on the participation index Calculating popularity score

[0206] .

[0207] In one embodiment, the course difficulty module 50 is further used to obtain historical performance information in the candidate courses and determine the course difficulty index DI according to the historical performance information.

[0208]

[0209] in The average grade of the course; the weighted average method is used to combine historical grade information to calculate the learning ability index LAI

[0210]

[0211] in is the grade of the i-th course, is the weight of the i-th course; according to the course difficulty index DI and learning ability index LAI Calculate the Adaptive Difficulty Index (ADI) for each course

[0212]

[0213] in is the weight parameter.

[0214] In one embodiment, the satisfaction module 60 is also used to obtain comment area information in the candidate course information, the comment area information includes text information and non-text information; determine a word set in the text information, and determine a tag set in the word set according to a part-of-speech analysis strategy; calculate a text sentiment score in the tag set; determine non-text confidence information and sentiment classification ratio in the non-text information; determine a non-text sentiment score based on the non-text confidence information and the sentiment classification ratio; and determine the satisfaction level based on the text sentiment score and the non-text sentiment score.

[0215] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of the present application. In practical applications, technicians in this field can select part or all of it according to actual needs to achieve the purpose of the present embodiment, and no limitation is made here.

[0216] In addition, for technical details not fully described in this embodiment, please refer to the method for course recommendation based on multidimensional data analysis provided in any embodiment of the present application, which will not be repeated here.

[0217] In addition, it should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0218] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0219] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as a read-only memory (ROM) / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods of each embodiment of the present application.

[0220] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A course recommendation method based on multidimensional data analysis, characterized in that: include: Obtain target history data corresponding to the target user, and construct a user-course matrix based on the target history data to determine candidate course information; Determining course suitability based on the candidate course information; Obtaining course selection data information and constructing a hypergraph, and determining course compatibility through the hypergraph; Determine the number of students enrolled and the average completion rate in the candidate course information to generate a participation index, and generate popularity information according to the participation index; Determine the average score and user basic information corresponding to the candidate courses to generate the course difficulty; Obtaining comment area information in the candidate course information to determine the satisfaction level; A comprehensive recommendation score is generated according to the course fit, the course compatibility, the popularity information, the course difficulty and the satisfaction level.

2. The course recommendation method based on multidimensional data analysis according to claim 1, characterized in that: The step of obtaining target history data corresponding to the target user and constructing a user-course matrix according to the target history data to determine candidate course information includes: Obtain target history data corresponding to the target user, and construct a user-course learning matrix based on the target history data; Determine similar users in the user-course learning matrix using a user-based collaborative filtering recommendation algorithm; Obtain the courses that the target user has learned and extract the subject words of the courses that have been learned through the LDA topic model to generate course features; Candidate course information is determined according to the similar users and the course characteristics.

3. The course recommendation method based on multidimensional data analysis according to claim 2 is characterized in that: The step of determining candidate course information according to the similar users and the course characteristics comprises: Determine, according to the similar users, a set of target courses that the similar users have studied but the target user has not studied as a user similarity calculation result; Obtaining the learned course information of the target user, and determining the course similarity calculation result according to the learned course information and the course characteristics; The candidate course information is generated according to the user similarity calculation result and the course similarity calculation result.

4. The course recommendation method based on multidimensional data analysis according to claim 1, characterized in that: The step of obtaining course selection data information and constructing a hypergraph, and determining course compatibility through the hypergraph includes: Obtaining course selection data information, and determining node information and hyperedge information according to the course selection data information; Constructing a hypergraph according to the node information and the hyperedge information; converting the hypergraph into an incidence matrix; identifying course combinations in the association matrix using a community detection algorithm; Calculate the support, confidence and lift corresponding to the course combination; The course compatibility is determined according to the support, the confidence, and the improvement.

5. The course recommendation method based on multidimensional data analysis according to claim 2, characterized in that: The step of determining the number of students enrolled and the average completion rate in the candidate course information to generate a participation index, and generating popularity information according to the participation index comprises: Obtain historical data of candidate courses, and determine the number of students enrolled in each course from the historical data and the average completion rate per course ; Determine the number of students enrolled in each of the candidate courses according to the similar users ; According to the number of students enrolled and the average completion rate Calculating the participation index ； According to the participation index Calculating popularity score 。 6. The course recommendation method based on multidimensional data analysis according to claim 1, characterized in that: The step of determining the average scores and user basic information corresponding to the candidate courses to generate the course difficulty includes: Obtain historical performance information in the candidate course, and determine the course difficulty index DI based on the historical performance information in is the course average grade; The learning ability index is calculated by combining the historical performance information using the weighted average method. LAI in is the grade of the i-th course, is the weight of the i-th course; According to the course difficulty index DI and the learning ability index LAI Calculate the difficulty index for each course ADI in is the weight parameter.

7. The course recommendation method based on multidimensional data analysis according to claim 1, characterized in that: The step of obtaining the comment area information in the candidate course information to determine the satisfaction level includes: Acquire comment area information in the candidate course information, wherein the comment area information includes text information and non-text information; Determining a word set in the text information, and determining a tag set in the word set according to a part-of-speech analysis strategy; Calculating text sentiment scores in the tag set; Determining non-text confidence information and sentiment classification ratio in the non-text information; Determining a non-text sentiment score according to the non-text confidence information and the sentiment classification ratio; A satisfaction level is determined based on the text sentiment score and the non-text sentiment score.

8. A course recommendation system based on multidimensional data analysis, characterized in that: The course recommendation system based on multidimensional data analysis includes: A candidate course information module is used to obtain target history data corresponding to a target user, and to construct a user-course matrix based on the target history data to determine candidate course information; A course suitability module, used to determine course suitability according to the candidate course information; A course compatibility module is used to obtain course selection data information and construct a hypergraph, and determine the course compatibility through the hypergraph; A popularity information module, used for determining the number of students enrolled and the average completion rate in the candidate course information to generate a participation index, and generating popularity information according to the participation index; A course difficulty module, used to determine the average scores and user basic information corresponding to the candidate courses to generate the course difficulty; A satisfaction degree module, used for obtaining comment area information in the candidate course information to determine the satisfaction degree; A comprehensive recommendation scoring module is used to generate a comprehensive recommendation score based on the course fit, the course compatibility, the popularity information, the course difficulty and the satisfaction level.

9. A computer device, characterized in that: The device comprises: a memory and a processor, and when the processor runs the computer instructions stored in the memory, the processor executes the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The method comprises instructions, which, when executed on a computer, cause the computer to execute the method according to any one of claims 1 to 7.

Images