Education course recommendation system

Abstract

The invention discloses an education curriculum recommendation system, and relates to the technical field of education, and the system comprises a curriculum feature extraction module, a student feature analysis module, a recommendation algorithm module and a recommendation curriculum generation module. Interest preference consideration concentration degree and diversity, cognitive level attention improvement potential and thought attainment measurement maturity provide a basis for precise course matching and stimulate student enthusiasm, course feature extraction newly adds thought depth and timeliness evaluation, it is ensured that content has connotation and fits reality, a recommendation algorithm integrates multiple factors, and the recommendation algorithm has a good recommendation effect. The correlation and the discrete degree are considered, courses are accurately screened, and the matching precision is improved.

Description

Technical Field

[0001] The present invention relates to the field of educational technology, and specifically provides an educational course recommendation system. Background Art

[0002] In the traditional field of educational course recommendation, there are many significant disadvantages and deficiencies, seriously restricting the recommendation effect and students' learning experience. First, the learning ability assessment is one-sided, relying solely on simple conventional index weighting to measure learning ability. For example, simply using exam scores or learning time as the main basis completely ignores key factors such as the stability of learning ability. It is difficult to accurately match courses suitable for students' ability levels, which may lead to a mismatch between the course difficulty and the actual abilities of students.

[0003] Secondly, the traditional field of educational course recommendation only focuses on students' superficial preferences for obvious course themes or teaching methods, fails to comprehensively and deeply analyze the concentration and diversity of interests, and cannot fully understand the diversity and depth of students' interests. This makes the recommended courses too limited, unable to meet the diverse interest needs of students, restricts the expansion of students' learning horizons and the continuous stimulation of learning enthusiasm, and the assessment of cognitive level is limited to a simple judgment of knowledge mastery, lacking consideration of improvement potential, ignoring the potential for thinking development and knowledge transfer ability, and it is difficult to effectively promote the improvement of ideological qualities and cognitive expansion in course recommendation.

[0004] Finally, in the extraction of traditional course features, the dimension of the ideological depth of the course is ignored, only focusing on superficial aspects such as course themes and teaching objectives. This makes it impossible for students to access course content with deep connotations and values, restricts the deepening of ideological understanding and the improvement of theoretical level, and is not conducive to cultivating critical thinking and in-depth analysis ability.

[0005] Therefore, we propose an educational course recommendation system to solve the above problems. Summary of the Invention

[0006] The purpose of the present invention is to make up for the deficiencies of the prior art and provide an educational course recommendation system. It can comprehensively and deeply analyze students' characteristics through the student characteristic analysis module, consider both conventional and stability in learning ability assessment, take into account the concentration and diversity of interest preferences, focus on improvement potential in cognitive level, measure the maturity of ideological qualities, provide a basis for accurate course matching and stimulate students' enthusiasm. In course feature extraction, new assessments of ideological depth and timeliness are added to ensure that the content has connotations and is practical. The recommendation algorithm synthesizes multiple factors, considers relevance and dispersion degree, etc., accurately screens courses, and improves the matching accuracy.

[0007] To solve the above technical problems, the present invention provides the following technical solution: an educational course recommendation system, which includes: a course feature extraction module, a student characteristic analysis module, a recommendation algorithm module, and a recommended course generation module;

[0008] Course Feature Extraction Module: It is used to extract various feature information of educational courses. The feature information includes course themes, teaching objectives, teaching methods, case content, course difficulty levels, and the dimension of the depth of thought in the course. The dimension of the depth of thought in the course is determined by analyzing the theoretical depth, philosophical thinking level involved in the course content, and the potential influence on the shaping of students' values. Let the depth of thought evaluation value be D i , and its calculation formula is: where z is the number of factors related to the depth of thought, w dv is the weight of the v-th factor related to the depth of thought, which is determined through expert evaluation and the analysis of a large number of excellent educational courses. X iv is the quantization value of course i on the v-th factor related to the depth of thought, including but not limited to the number of citations of classical theories included in the course and the number of questions guiding students to think deeply;

[0009] Student Feature Analysis Module: Through students' learning history data, exam scores, classroom performance, and online learning behavior data, the following algorithm is used to analyze students' learning ability, interest preferences, cognitive level, and ideological quality status:

[0010] Let the student learning ability evaluation value be C a , the interest preference evaluation value be C i , the cognitive level evaluation value be C c , and the ideological quality evaluation value be C s ;

[0011] Learning Ability Evaluation Formula: where n is the number of indicators related to learning ability, w aj is the weight of the j-th learning ability indicator, which is determined through the statistical analysis of a large amount of student data and the training of machine learning models. S aj is the score of the student on the j-th learning ability indicator, including but not limited to the indicator scores of learning time and homework completion quality. λ is a regulation coefficient, which is determined according to the dispersion degree of learning ability data and is used to measure the stability of students' learning ability, is the average value of the learning ability indicator scores;

[0012] Interest Preference Evaluation Formula: where m is the number of dimensions related to interest preferences, w ik is the weight of the k-th interest preference dimension, which is determined based on students' participation and feedback data on different course themes and teaching methods. I ik$x_{k}$ is the performance value of the student in the $k$-th interest preference dimension, including but not limited to the browsing duration of specific topic courses and the enthusiasm for participating in discussions. $\mu$ is a correction coefficient, determined according to the changing trend of the student's interest preference, used to adjust the concentration degree of the interest preference. The second term is used to measure the diversity of the student's interest preference;

[0013] Cognitive level evaluation formula: where $p$ is the number of cognitive level evaluation factors, $w$ cl is the weight of the $l$-th cognitive level factor, determined based on the student's scores on different knowledge points in the exam and the accuracy rate of answering questions in class. $E$ cl is the evaluation value of the student in the $l$-th cognitive level factor, including but not limited to the mastery of basic knowledge and advanced theories. $\xi$ is a regulation factor, determined according to the difficulty of improving the cognitive level. The second term is used to reflect the comprehensive improvement potential of the student's cognitive level;

[0014] Thought quality evaluation formula: where $r$ is the number of thought quality related indicators, $w$ sq is the weight of the $q$-th thought quality indicator, determined through the student's performance in relevant activities and the analysis of views on social hot issues. $Y$ sq is the score of the student in the $q$-th thought quality indicator, including but not limited to the enthusiasm for participating in public welfare activities. $Y$ sq is a correction parameter, determined according to the development stage of the student's thought quality, used to adjust the focus of thought quality evaluation. The second term is used to measure the maturity of the student's thought quality;

[0015] Recommendation algorithm module: Based on course characteristics and student characteristics, the following recommendation algorithm is used to calculate the matching degree of the recommended courses:

[0016] Let the matching degree between course $i$ and the student be $M$ i , the course feature vector be $F$ i =(f i1 , f i2 , …, f iq ), and the student feature vector be $U=(u 1 , u 2 , …, u r ), where $q$ is the number of course characteristics and $r$ is the number of student characteristics;

[0017] Matching degree calculation formula: where $\alpha$, $\beta$, $\gamma$, $\delta$ are regulation parameters, determined through experimental comparison and model optimization to balance the contributions of different factors to the matching degree. $w$ ft is the weight of course feature $t$, determined according to the importance and relevance of the course. $w$ usis the weight of student feature s, determined according to the influence degree of the student feature on the learning effect, f it is the value of course i on the t-th course feature, u t is the value of the student on the t-th feature, |u s -f is | represents the absolute value of the difference between the student feature s and the course feature s, -f t is the average value of the course feature t, is the average value of the student feature s. The third term is used to measure the correlation degree between the course feature and the student feature, and the fourth term is used to consider the influence of the dispersion degree of the course feature and the student feature on the matching degree;

[0018] Recommended course generation module: According to the matching degree result, screen out the courses with a matching degree higher than the preset threshold, generate a recommended course list and push it to the students.

[0019] Furthermore, the course feature extraction module also includes extracting the timeliness feature of the course. By analyzing the correlation degree between the course content and the current social hotspots, policies and regulations, as well as the course update cycle, set the timeliness evaluation index T. The specific calculation method is: where g is the number of timeliness-related factors, w th is the weight of the h-th timeliness factor, determined according to the degree of influence of the course by social hotspots in historical data and the importance of policy and regulation updates to the course, H h is the performance value of the course on the h-th timeliness factor, including but not limited to the number of cases involving current hot events in the course and the depth of interpretation of the latest policies and regulations, T u is the update cycle of the course, in days. τ is an adjustment factor, determined according to the importance degree of the course timeliness, used to adjust the influence weight of the update cycle on the timeliness evaluation.

[0020] Even further, when analyzing the learning historical data of the student, the student feature analysis module further considers the learning time distribution feature and the learning location preference feature of the student. By calculating the learning activity A of the student in different time periods d and the learning efficiency of different learning locations to improve the learning ability evaluation. The learning activity formula is: where N d is the number of learning behaviors of the student in the time period, N t is the total number of learning behaviors. The learning efficiency formula is: where m is the number of learning location-related factors, w li is the weight of the i-th learning location factor, determined by analyzing the learning achievements and learning behaviors of the student in different locations, S liLet \(S_i\) be the score of the student on the \(i\)-th learning location factor, including but not limited to the learning concentration in the library and the participation in the classroom. Incorporate learning activity and learning efficiency as supplementary factors in the learning ability assessment formula \(C\). a That is where \(\varphi\) is the comprehensive adjustment coefficient, determined according to the influence degrees of learning activity and learning efficiency on learning ability, to more comprehensively reflect the influence of students' learning habits and environment on learning ability.

[0021] Furthermore, in the recommendation algorithm module, when calculating the matching degree between a course and a student, introduce the interactivity feature \(I\) of the course c . The interactivity feature includes the setting of group discussion sessions in the course, the frequency of teacher-student interaction, and the activity of the online interaction platform. The influence of the interactivity feature on the matching degree is reflected by the following formula: \(M'\) i = \(M\) i ×(1 + \(\kappa I\)) c , where \(M'\) i is the matching degree considering interactivity, \(\kappa\) is the interactivity adjustment coefficient, determined according to experimental data, to reflect the influence of course interactivity on students' learning experience and participation. The higher the course interactivity, the relatively increased weight in the recommendation, encouraging students to participate in courses with strong interactivity and improving learning effects.

[0022] Furthermore, when the recommended course generation module generates a list of recommended courses, sort them in descending order of the matching degree, and highlight several courses with the top rankings. At the same time, provide a detailed course introduction and recommendation reasons for each recommended course. The recommendation reasons include the fit points of the course with the students' interest preferences, the help in improving the students' cognitive level, the relevance to the current learning goals, and the characteristic advantages of the course.

[0023] Furthermore, the system also includes a user feedback collection module for collecting students' feedback information on the recommended courses. The feedback information includes course satisfaction ratings, self-evaluations of learning effects, and opinions and suggestions on course content and recommendation accuracy. After quantifying the feedback information, it is used to adjust the parameters and weights in the recommendation algorithm. The specific adjustment methods are as follows

[0024] For the learning ability assessment weight \(w\) aj , adjust it according to the correlation analysis between the students' self-evaluations of learning effects and the learning ability assessment values. If the actual learning effects of the students are higher than expected and some of the learning ability assessment values are lower, appropriately increase the corresponding weights. For the interest preference assessment weight \(w\) ik, adjust according to the relationship between the student's satisfaction score for the course and the degree of fit of interest preferences. If the student has a high satisfaction with a certain type of course but the weight of the relevant dimension in the interest preference assessment is low, then increase the weight of this dimension;

[0025] For the weight w of course characteristics ft and the weight w of student characteristics us , according to the students' opinions on the course content and the accuracy of recommendations, combined with data analysis, adjust the weights of the parts that do not match the students' feedback to continuously optimize the accuracy and adaptability of the recommendation system. At the same time, according to the students' feedback information, dynamically adjust the matching threshold to improve the precision of recommendations and the students' satisfaction.

[0026] Furthermore, the system is set with a function to switch personalized recommendation modes. Students can switch between the precise recommendation mode and the exploratory recommendation mode according to their own needs. In the precise recommendation mode, the recommendation algorithm focuses more on matching based on the students' existing characteristics and historical data to provide courses that highly match the students' current interests and abilities;

[0027] In the exploratory recommendation mode, the system will appropriately lower the matching threshold, introduce some courses with a certain degree of novelty and challenge, encourage students to expand their learning fields and explore unknown knowledge, and at the same time clearly mark the novelty and challenge indicators of the courses in the recommendation list to help students better understand the course characteristics. The novelty indicator is determined by calculating the similarity between the course content and the courses the students have already studied. Let the novelty indicator be N, and the calculation formula is:

[0028] where u it is the relevant characteristic value that the student already has on the course characteristic t. The challenge indicator is evaluated based on the difference between the difficulty level of the course and the students' cognitive level. Let the challenge indicator be C h , and the calculation formula is where C ci is the current cognitive level evaluation value of the student, C ti is the difficulty level evaluation value of the course, C max and C min are respectively the maximum and minimum values of the course difficulty levels in the system.

[0029] Furthermore, the course feature extraction module and the student feature analysis module adopt a distributed computing architecture. The specific distributed computing strategy is as follows: divide the course data and student data into multiple computing nodes according to the data volume and relevance. Each computing node parallelly processes the feature extraction and analysis tasks of part of the data, and then aggregates the results to the master node for integration. When dividing the data, the hash algorithm is used to ensure the balanced distribution of data. At the same time, considering the relevance of the data, the relevant data is allocated to the same computing node or adjacent computing nodes to reduce data transmission and communication overhead, and improve the operation efficiency and response speed of the entire system.

[0030] Furthermore, the system has an interface with the school's teaching management system, which can obtain the information of students' course selection, teaching progress arrangement, and the school's education and teaching plan in real time, and integrate this information into the recommendation algorithm, so that the recommended courses conform to the school's overall teaching plan and the actual learning process of students. At the same time, according to the school's teaching focus and training objectives, adjust the weight parameters in the recommendation algorithm to highlight the course recommendations that meet the school's education direction. Let the weight adjustment coefficient of the school's key courses be ω s , when the course belongs to the education direction that the school focuses on, w f t = w ft ×(1 + ω s ), where w f t is the weight of the course feature t.

[0031] Furthermore, the system is equipped with a security and privacy protection module, which uses encryption technology to encrypt and store and transmit students' learning data and personal information to ensure the security and confidentiality of the data. During the data usage process, follow strict privacy policies, anonymize the data, remove sensitive information that can directly identify students' identities, and only retain the necessary feature information for the recommendation algorithm calculation. At the same time, establish a data access permission management system to restrict unauthorized personnel from accessing and using students' data, protect the privacy rights and interests of students from infringement, and use blockchain technology to store and trace the operation records of the data to ensure the integrity and auditability of the data, and prevent the data from being tampered with or misused.

[0032] Compared with the prior art, the educational course recommendation system has the following beneficial effects:

[0033] 1. The present invention introduces a student feature analysis module, comprehensively and deeply considering students' learning ability, interest preference, cognitive level, and ideological quality. The learning ability evaluation formula not only includes the weighting of conventional indicators but also introduces a regulation coefficient to measure stability, which can accurately reflect the true state of students' learning ability and provide a strong basis for matching courses suitable for their abilities. The interest preference evaluation formula takes into account both the concentration and diversity of interests to ensure that the recommended courses highly match the students' interest points, thereby stimulating students' learning enthusiasm and initiative. The part reflecting the improvement potential in the cognitive level evaluation formula and the setting measuring maturity in the ideological quality evaluation formula further refine the understanding of students, enabling the recommended courses to not only meet the current cognitive level but also contribute to the improvement of ideological quality and the expansion of cognition.

[0034] 2. The course feature extraction module of the present invention adds the ideological depth dimension and timeliness evaluation on the basis of traditional features. The evaluation of the ideological depth dimension ensures that students can access more profound and valuable course content, which helps to deepen students' ideological understanding and theoretical level. The timeliness evaluation guarantees that the course content is closely combined with current social hotspots and policies and regulations, making the knowledge learned by students more practically applicable and timely. The matching degree calculation formula in the recommendation algorithm module synthesizes various factors, comprehensively considering various relationships between courses and student features, such as the degree of relevance and dispersion, etc., and can accurately screen out the most suitable courses for students, improving the matching accuracy between courses and students' needs.

[0035] Other advantages, objectives, and features of the present invention will be described to some extent in the subsequent specification, and to some extent, will be obvious to those skilled in the art based on the study of the following text, or can be taught from the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0037] Figure 1 It is a flowchart of an educational course recommendation system. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0038] The technical solutions in the embodiments of the present invention will be described clearly and completely below. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0039] Embodiment 1: College Education Course Recommendation

[0040] In a comprehensive university, in order to improve the teaching effect of education courses and the learning enthusiasm of students, this education course recommendation system is introduced. The university has rich education course resources, covering multiple courses such as Course A, Course B, and Course C. At the same time, the student group has diverse learning needs and backgrounds.

[0041] Course Feature Parameter and Weight Setting:

[0042] Course Theme: Divided into multiple themes such as philosophy, law, and history, and classified and assigned values according to the core theme of the course content. For example, Course A mainly focuses on the philosophy theme and is assigned a value of 0.8 (assuming the weight range is 0 - 1);

[0043] Teaching Objectives: Divided into objectives such as knowledge imparting, ability cultivation, and value shaping, and the weights are allocated according to the emphasis on each objective in the course syllabus. For example, for a course that emphasizes knowledge imparting and value shaping, the knowledge imparting weight is set to 0.4, the value shaping weight is set to 0.4, and the ability cultivation weight is set to 0.2;

[0044] Teaching Methods: Include lecture method, discussion method, case analysis method, practical teaching method, etc., and the weights are set according to the combination of teaching methods actually adopted in the course. For example, for a course mainly using the lecture method and case analysis method, the lecture method weight is 0.5, the case analysis method weight is 0.3, and the total weight of other methods is 0.2;

[0045] Case Content: Evaluated according to the timeliness, typicality, and relevance of the case. Cases with strong timeliness, typicality, and high relevance to the course theme are assigned higher values. Assuming the weight is 0.3, and general cases are assigned values of 0.1 - 0.2;

[0046] Course Difficulty Level: Divided into primary, intermediate, and advanced levels, and divided according to the knowledge depth, theoretical difficulty, etc. of the course. The primary course difficulty level is assigned a value of 0.3, the intermediate level is 0.5, and the advanced level is 0.7;

[0047] Thought Depth Dimension: Through the evaluation of experts on the theoretical depth, philosophical thinking level, and potential influence on students' value shaping in the course. For example, for a certain course that performs outstandingly in this regard, the thought depth evaluation value D i of the relevant weight w dvThe sum is 0.6. The weights of each factor are determined according to specific analysis. For example, the weight of the number of citations of classical theories is 0.2, and the weight of the number of deeply considered questions is 0.3, etc. Timeliness characteristics: Consider the relevance of the course content to current social hotspots, policies and regulations, as well as the course update cycle. For courses that are closely related to recent hotspots and have a short update cycle, the weight w of timeliness-related factors th The sum is 0.5. Among them, the weight of the number of hot event cases is 0.2, the weight of the depth of policy and regulation interpretation is 0.3, and the weight τ of the reciprocal of the update cycle is set to 0.1 according to the importance of the course timeliness (assuming the course update cycle is in months, and the value is larger when the update cycle is short ).

[0048] Student characteristic parameters and weight settings: Learning ability: Learning ability-related indicators include learning time, quality of homework completion, improvement rate of exam scores, etc. The weight w of learning time aj is set to 0.3 according to statistical analysis (such as the proportion of daily learning duration), the weight of the quality of homework completion is 0.4 (evaluated according to the homework score), the weight of the improvement rate of exam scores is 0.3, and the adjustment coefficient λ is set to 0.2 according to the analysis of the dispersion degree of learning ability data (assuming that the value of λ increases appropriately when the dispersion degree is large to emphasize stability);

[0049] Interest preference: The interest preference dimension includes preferences for different course topics, teaching methods, etc. For students interested in the philosophy topic, the weight w of this dimension ik is set to 0.3, the weight of being interested in the discussion teaching method is set to 0.2, and the sum of the weights of other dimensions is 0.5. The correction coefficient μ is set to 0.1 according to the change trend of students' interest preferences (if students' interests gradually converge, the value of μ decreases);

[0050] Cognitive level: Cognitive level evaluation factors include the degree of basic knowledge mastery, understanding of advanced theories, application ability of knowledge, etc. The weight w of the degree of basic knowledge mastery cl is set to 0.4, the weight of the understanding of advanced theories is 0.3, the weight of the application ability of knowledge is 0.3, and the adjustment factor ξ is set to 0.15 according to the difficulty of cognitive level improvement (the value of ξ increases appropriately when the improvement difficulty is large to reflect potential);

[0051] Thought quality: Thought quality indicators include the enthusiasm for participating in public welfare activities, the depth of understanding of Course A, ideological awareness, etc. The weight w of the enthusiasm for participating in public welfare activities sq is set to 0.3, the weight of the depth of understanding of Course A is 0.4, the weight of ideological awareness is 0.3, and the correction parameter ω is set to 0.1 according to the development stage of students' thought quality (the value of ω is adjusted when the development stage is relatively high to focus on the measurement of maturity);

[0052] Learning time distribution characteristics: Divide a day into different time periods, such as morning, afternoon, and evening. Calculate the learning activity A based on the proportion of the number of learning behaviors of students in each time period. d , assuming that the learning activity is relatively high in the morning, and its weight accounts for 0.6 in the supplementary factors for learning ability evaluation, 0.3 in the afternoon, and 0.1 in the evening. In the learning location preference characteristics, the weight w li of the learning efficiency in the library is set to 0.4, the weight of the learning efficiency in the classroom is 0.3, and the total weight of other learning locations is 0.3. The comprehensive adjustment coefficient is set to 0.15 according to the influence degree of learning activity and learning efficiency on learning ability.

[0053] Recommendation algorithm parameter settings:

[0054] The adjustment parameters α, β, γ, δ are determined through experimental comparison and model optimization. Assume they are 0.3, 0.2, 0.2, 0.1 respectively. The weight w ft of the course characteristics and the weight w us of the student characteristics are comprehensively determined according to the importance and relevance of the course and student characteristics. For example, if the relevance between the course theme and the student's interest preference is high, the weights of the course theme characteristics and the corresponding student interest preference characteristics are appropriately increased.

[0055] Implementation process:

[0056] Course feature extraction: The course feature extraction module analyzes the educational courses offered by the school. For example, for Course A, determine that its course theme is philosophy (assigned a value of 0.8), the teaching objectives are mainly knowledge imparting (weight 0.4) and value shaping (weight 0.4), the teaching methods are lecture method (weight 0.5) and case analysis method (weight 0.3), the case content has high timeliness and relevance (assigned a value of 0.3), the course difficulty level is intermediate (assigned a value of 0.5), and the comprehensive evaluation value D is obtained through expert evaluation of various factors in the dimension of ideological depth. i , The timeliness feature is evaluated according to the degree of association with current social hotspots (such as the number of cases involving current hot philosophical issues) and the update cycle (assuming the last update was half a year ago, and the weight τ of the reciprocal of the update cycle is calculated and incorporated into the evaluation).

[0057] Student characteristic analysis:

[0058] The student characteristic analysis module collects the learning history data, exam scores, classroom performance, and online learning behavior data of students. Taking a certain student as an example, by counting his learning time, calculate the learning activity A in different time periods. d , assuming that the learning activity in the morning is 70%, 50% in the afternoon, and 30% in the evening. According to his learning performance and behavior performance in different learning locations such as the library and the classroom, evaluate the learning efficiency Ei , the learning efficiency in the library is relatively high at 80%, and in the classroom it is 60%. At the same time, the learning ability evaluation value C is calculated based on the quality of homework completion, the improvement rate of exam scores, etc. a , the interest preference evaluation value C is determined by combining the participation in different course topics and teaching methods, etc. i , the cognitive level evaluation value C is determined based on the mastery of different knowledge points in the exam scores. c , the ideological quality evaluation value C is evaluated through the participation in public welfare activities. s ;

[0059] Recommendation algorithm calculation:

[0060] The recommendation algorithm module calculates the matching degree M between each course and the student using the matching degree calculation formula based on the course feature vector F i and the student feature vector U. For example, for a basic law course, the course feature vector is F i =(f i i1 , f i2 , …, f iq ), and the student feature vector is U=(u 1 , u 2 , …, u r ). Substitute each parameter and weight into the formula: for calculation, where α = 0.3, β = 0.2, γ = 0.2, δ = 0.1, w f t and w u s are substituted according to the weight values determined previously, f it and u t are the values of the course and the student on each feature respectively, and are the average values of the course features and the student features;

[0061] Recommended course generation:

[0062] ​The recommended course generation module filters out courses with a matching degree higher than a preset threshold (assumed to be 0.6) based on the matching degree results, generates a list of recommended courses and pushes them to students. For example, the recommended list may include courses that highly match the student's interest preferences, are helpful for improving their cognitive level, and meet the current learning goals. For instance, according to the student's interest in philosophical themes and discussion-based teaching methods, as well as their current cognitive level and ideological accomplishment status, a seminar course mainly focused on philosophical case analysis is recommended. At the same time, the points of fit between the course and the student's interest preferences (such as including rich philosophical case discussions), the help for improving the cognitive level (helpful for in-depth understanding of philosophical theories and cultivating analytical abilities), and the relevance to the current learning goals (meeting the overall goals of school education and the student's stage learning needs) are detailed in the recommendation reasons.

[0063] Example Two: Course Recommendation on an Online Education Platform

[0064] An online education platform provides diverse educational courses for a large number of online learners, including video courses, online lectures, interactive seminar courses, etc. The platform hopes to improve users' learning experience and learning effect through accurate course recommendations, and attract more users to participate in learning.

[0065] Course Feature Parameter and Weight Setting:

[0066] Course Theme: Classified according to themes such as B Course Theory, B Course Cultivation, B Course Inheritance, and Social Hotspot Analysis. For example, the weight of courses on the B Course Theory theme is set to 0.3, which is determined according to the distribution of platform course resources and users' attention to each theme.

[0067] Teaching Objectives: With knowledge popularization, thinking expansion, and behavior guidance as the main objectives, the weight of the knowledge popularization objective is set to 0.3, the weight of thinking expansion is 0.4, and the weight of behavior guidance is 0.3. The weight distribution is determined according to the original intention of course design and the educational concept of the platform.

[0068] Teaching Methods: Include video explanations, online interactions, group assignments, case analyses, etc. The weight of video explanations is 0.4, the weight of online interactions is 0.3, and the total weight of group assignments and case analyses is 0.3. The weights are determined according to the effectiveness of teaching methods and the platform's technical support capabilities.

[0069] Case Content: Evaluated according to the novelty, representativeness, and educational significance of cases. The weight of novel and representative cases is 0.4, and the weight of general cases is 0.2 - 0.3.

[0070] Course Difficulty Level: Divided into basic, advanced, and advanced levels. The weight of the basic course difficulty level is 0.3, the advanced level is 0.5, and the advanced level is 0.7, which is divided according to the knowledge depth and learning requirements of the course.

[0071] Depth dimension of thought: By conducting in-depth analysis of the course content and expert evaluation, the weights of various factors are determined. For example, the weight of the depth of interpretation of classic works is 0.3, the weight of critical thinking on social trends is 0.4, and the sum of the weights of other factors is 0.3;

[0072] Timeliness characteristics: Considering the degree of closeness of the course content to current hot topics and the timeliness of updates, for courses that are closely related to popular social events and have been recently updated, the sum of the weights of timeliness-related factors is 0.6, among which the weight of the association with hot topics is 0.3, and the weight of update timeliness is 0.3. The update cycle is in days, and the corresponding weights of courses with a short update cycle are adjusted more significantly.

[0073] Student characteristic parameters and weight settings:

[0074] Learning ability: Evaluated through indicators such as the user's learning duration, course completion progress, homework and test scores, etc. The weight of learning duration is 0.3, the weight of course completion progress is 0.3, and the weight of homework and test scores is 0.4. The adjustment coefficient is determined according to the stability of the user's learning data, assumed to be 0.15;

[0075] Interest preference: Analyzed based on the user's browsing history, course collections, comments and interaction behaviors, etc., and weight distribution is carried out for the interest preferences for different topics and teaching methods. For example, the weight of being interested in the topic of cultural inheritance is 0.3, the weight of being interested in interactive teaching methods is 0.2, and the sum of the weights of other dimensions is 0.5. The correction coefficient is determined according to the trend of the user's interest changes, assumed to be 0.12;

[0076] Cognitive level: Based on the user's learning performance and test results on the platform, evaluate their mastery of basic knowledge, intermediate knowledge and advanced knowledge, and set the weights to 0.3, 0.4, and 0.3 respectively. The adjustment factor is determined according to the difficulty of improving the cognitive level, assumed to be 0.18;

[0077] Thought quality: Evaluated through the activity of the user's participation in relevant course discussions, the depth and correctness of viewpoints, etc. The weight of the activity of participating in discussions is 0.3, the weight of the depth and correctness of viewpoints is 0.4, and the weight of other factors is 0.3. The correction parameter is determined according to the development stage of the user's thought quality, assumed to be 0.13;

[0078] Learning time distribution characteristics: Based on the frequency and duration of users logging in to the platform for learning at different time periods, determine the learning activity weights for each time period. For example, the learning activity is relatively high at night, and the weight is set to 0.4; during the day's working hours, the weight is relatively low at 0.2; for other time periods, the weight is 0.4. The learning location preference characteristics are mainly reflected in the stability of the devices and network environment used in the online platform. Assume that the learning efficiency weight for using a computer is 0.5, the weight for using a mobile device is 0.3, and the weight increases by 0.1 when the network environment is stable. The comprehensive adjustment coefficient is set to 0.12 according to the actual situation.

[0079] Recommendation algorithm parameter settings:

[0080] The adjustment parameters α, β, γ, δ are determined through experimental comparison of a large amount of user data on the platform and model optimization. Assume they are 0.25, 0.25, 0.3, 0.2 respectively. The course feature weight w ft and the student feature weight w us are comprehensively determined according to the characteristics of the platform courses and the user group characteristics. For example, for popular course topics and characteristics that users generally pay attention to, the corresponding weights are appropriately increased.

[0081] Implementation process:

[0082] Course feature extraction: The course feature extraction module conducts a comprehensive analysis of the courses on the platform. For example, for a video course on Course B, determine that its course theme is Course B cultivation (assigned a value of 0.3), the teaching objectives are mainly thinking expansion (weight 0.4) and behavior guidance (weight 0.3), the teaching methods are mainly video lectures (weight 0.4) and online interaction (weight 0.3), the case content has high novelty and educational significance (assigned a value of 0.4), the course difficulty level is advanced (assigned a value of 0.5), the ideological depth dimension obtains a comprehensive value through the evaluation of the interpretation of classic theories and the analysis of social reality problems in the course, and the timeliness feature is evaluated according to the degree of association with current hot topics and the update cycle (assuming the last update was a month ago, calculate the corresponding weight according to the update cycle);

[0083] Student feature analysis:

[0084] The student feature analysis module collects the learning behavior data of users on the platform. Taking a certain user as an example, by counting their learning duration and course completion status on the platform, calculate the learning ability evaluation value C a , combine their browsing history (the number of times and duration of browsing courses on different topics), favorite collection behaviors, etc. to determine the interest preference evaluation value C i , evaluate the cognitive level evaluation value C according to their performance in platform tests and assignments c , and evaluate the ideological accomplishment evaluation value C through the frequency and quality of participating in discussions s, meanwhile, analyze the frequency and duration of logging in to the platform for learning in different time periods, and calculate the learning activity A d , and evaluate the learning efficiency E according to the situation of the device and network environment used i ;

[0085] Recommendation algorithm calculation:

[0086] The recommendation algorithm module calculates the matching degree M between each course and the user according to the course feature vector F i and the student feature vector U. For example, for an interactive seminar course on the inheritance of traditional Chinese culture, substitute the eigenvalue and corresponding weight of the course and the student into the matching degree calculation formula for calculation. Substitute the adjusted parameters and weights according to the previously set values. The specific calculation process is similar to that of Embodiment 1. By comprehensively considering multiple feature factors of the course and the student, obtain the matching degree result; i

[0087] Recommendation course generation:

[0088] The recommended course generation module filters out the courses with a matching degree higher than the preset threshold (assumed to be 0.55) according to the matching degree result, and generates a list of recommended courses to be pushed to the user. For example, for a user who is interested in the theme of cultural inheritance and has a certain cognitive level, a course that deeply explores the innovation of cultural inheritance in combination with historical cases is recommended. In the recommendation reason, explain the fit point between the course and the user's interest (such as rich cultural case analysis), the help for improving the cognitive level (guiding in-depth thinking about the theory and practice issues of cultural inheritance), and the relevance to the user's current learning goal (meeting the user's need to improve cultural quality and ideological understanding), and at the same time highlight the unique advantages of the course, such as inviting well-known experts for lectures and interactive guidance.

[0089] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and without departing from the spirit or basic characteristics of the present invention, the present invention can be implemented in other specific forms. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be included in the present invention. Any reference signs in the claims should not be regarded as limiting the claims involved.​

Claims

1. An educational course recommendation system, characterized in that: The system includes: a course feature extraction module, a student feature analysis module, a recommendation algorithm module and a recommended course generation module; Course feature extraction module: used to extract various feature information of educational courses, including course theme, teaching objectives, teaching methods, case content, course difficulty level and course thought depth dimension. The thought depth dimension of the course is determined by analyzing the theoretical depth, philosophical thinking level and potential influence on the shaping of students' values ​​involved in the course content. The thought depth evaluation value is D i , and its calculation formula is: Where z is the number of factors related to thought depth, w dv is the weight of the vth thought depth factor, determined through expert evaluation and analysis of a large number of excellent educational courses, X iv is the quantitative value of the vth depth of thought factor of course i, including but not limited to the number of classic theory references included in the course and the number of questions that guide students to think deeply; Student feature analysis module: Through students’ learning history data, test scores, classroom performance, and online learning behavior data, the following algorithms are used to analyze students’ learning ability, interest preferences, cognitive level, and ideological literacy: Assume that the student's learning ability evaluation value is C a , the interest preference evaluation value is C i , the cognitive level assessment value is C c , the ideological quality assessment value is C s ; Learning ability assessment formula: Where n is the number of learning ability related indicators, w aj is the weight of the jth learning ability indicator, which is determined by statistical analysis of a large amount of student data and machine learning model training. aj is the student's score on the jth learning ability indicator, including but not limited to the indicator scores of learning time and homework completion quality. λ is the adjustment coefficient, which is determined according to the discrete degree of learning ability data and is used to measure the stability of students' learning ability. is the average of the learning ability index scores; Interest preference evaluation formula: Where m is the number of interest preference related dimensions, w ik is the weight of the kth interest preference dimension, which is determined based on students’ participation and feedback data on different course topics and teaching methods. ik is the performance value of students in the kth interest preference dimension, including but not limited to the browsing time of specific subject courses and the enthusiasm for participating in discussions. μ is the correction coefficient, which is determined according to the changing trend of students' interest preferences and is used to adjust the concentration of interest preferences. The second item is used to measure the diversity of students' interest preferences. Cognitive level assessment formula: Where p is the number of cognitive level assessment factors, w cl is the weight of the lth cognitive level factor, which is determined based on the students’ scores of different knowledge points in the test and the accuracy of their answers to classroom questions. cl is the evaluation value of the student on the lth cognitive level factor, including but not limited to the mastery of basic knowledge and advanced theories, ξ is the adjustment factor, which is determined according to the difficulty of improving the cognitive level, and the second item is used to reflect the comprehensive improvement potential of the student's cognitive level; Ideological literacy assessment formula: Among them, r is the number of indicators related to ideological literacy, w sq is the weight of the qth ideological literacy indicator, which is determined by analyzing students’ performance in related activities and their views on social hot issues. sq is the student's score on the qth ideological literacy indicator, including but not limited to the enthusiasm for participating in public welfare activities, Y sq To correct the parameters, it is determined according to the development stage of students' ideological literacy and used to adjust the focus of ideological literacy assessment. The second item is used to measure the maturity of students' ideological literacy; Recommendation algorithm module: Based on the course characteristics and student characteristics, the following recommendation algorithm is used to calculate the matching degree of the recommended courses: Let the matching degree between course i and students be M i , the course feature vector is F i =(f i1 ,f i2 ,…,f iq ), the student characteristic vector is U=(u1,u2,…,u r ), where q is the number of course features and r is the number of student features; Matching degree calculation formula: Among them, α, β, γ, and δ are adjustment parameters, which are determined through experimental comparison and model optimization to balance the contribution of different factors to the matching degree. ft is the weight of course feature t, determined according to the importance and relevance of the course, w us is the weight of student feature s, which is determined according to the influence of student features on learning effect, f it is the value of course i on the tth course feature, u t is the value of the student on the tth feature, |u s -f is | represents the absolute value of the difference between student feature s and course feature s, is the average value of course feature t, is the average value of student characteristics s, the third item is used to measure the correlation between course characteristics and student characteristics, and the fourth item is used to consider the impact of the dispersion of course characteristics and student characteristics on the matching degree; Recommended course generation module: Based on the matching results, select courses with matching degrees higher than the preset threshold, generate a list of recommended courses and push it to students.

2. The educational course recommendation system according to claim 1, characterized in that: The course feature extraction module also includes extracting the timeliness feature of the course. By analyzing the relevance of the course content to current social hot spots, policies and regulations, and the course update cycle, the timeliness evaluation index T is set. The specific calculation method is: Where g is the number of time-related factors, w th is the weight of the hth timeliness factor, which is determined according to the degree to which the course is affected by social hot spots in historical data and the importance of policy and regulatory updates to the course. h is the performance value of the course on the hth timeliness factor, including but not limited to the number of cases involving current hot events in the course and the depth of interpretation of the latest policies and regulations. u is the update cycle of the course in days, and τ is the adjustment factor, which is determined according to the importance of course timeliness and is used to adjust the weight of the update cycle on the timeliness evaluation.

3. The educational course recommendation system according to claim 1, characterized in that: The student feature analysis module further considers the student's learning time distribution characteristics and learning location preference characteristics when analyzing the student's learning history data, and calculates the student's learning activity A in different time periods. d And the learning efficiency of different learning locations is used to improve the learning ability assessment. The learning activity formula is: Where N d is the number of learning behaviors of students in a time period, N t is the total number of learning behaviors, and the learning efficiency formula is: Where m is the number of factors related to the learning location, w li is the weight of the i-th learning location factor, which is determined by analyzing the learning performance and learning behavior of students in different locations. li The score of the student on the i-th learning location factor, including but not limited to the learning concentration in the library and the participation in the classroom, is integrated into the learning ability evaluation formula C by taking learning activity and learning efficiency as supplementary factors for learning ability evaluation. a In Among them, φ is the comprehensive adjustment coefficient, which is determined according to the influence of learning activity and learning efficiency on learning ability, so as to more comprehensively reflect the impact of students' learning habits and environment on learning ability.

4. The educational course recommendation system according to claim 1, characterized in that: In the recommendation algorithm module, the interactive feature of the course is introduced when calculating the matching degree between the course and the student. c ,The interactive features include the group discussion session setting in the course, the frequency of teacher-student interaction, and the activity of the online interactive platform. The impact of interactive features on matching is reflected by the following formula: M′ i =M i ×(1+κI c ), where M′ i To consider the matching degree after interactivity, κ is the interactivity adjustment coefficient, which is determined according to the experimental data to reflect the impact of course interactivity on students’ learning experience and participation. The higher the course interactivity, the greater its weight in the recommendation, which encourages students to participate in courses with strong interactivity and improve learning effects.

5. The educational course recommendation system according to claim 1, characterized in that: When generating a list of recommended courses, the recommended course generation module sorts the courses from high to low according to the degree of match, and highlights the top-ranked courses. At the same time, a detailed course introduction and recommendation reasons are provided for each recommended course. The recommendation reasons include the fit between the course and the students' interest preferences, the help in improving the students' cognitive level, the relevance to the current learning goals, and the characteristics and advantages of the course.

6. The educational course recommendation system according to claim 1, characterized in that: The system also includes a user feedback collection module, which is used to collect students' feedback on recommended courses. The feedback information includes course satisfaction scores, self-evaluation of learning effects, and opinions and suggestions on course content and recommendation accuracy. After quantification, the feedback information is used to adjust the parameters and weights in the recommendation algorithm. The specific adjustment method is as follows: For learning ability assessment weight w aj , according to the correlation analysis between students’ self-evaluation of learning effect and learning ability evaluation value, the corresponding weights of some indicators are appropriately increased if the students’ actual learning effect is higher than expected and the learning ability evaluation value is lower. ik , according to the relationship between students' course satisfaction scores and interest preference fit, adjustments are made. If students are highly satisfied with a certain type of course but the weight of the relevant dimension in the interest preference assessment is low, the weight of this dimension is increased; For the course feature weight w ft and student feature weights w us ,According to students’ opinions on course content and recommendation accuracy, combined with data analysis, the weights of the parts that are inconsistent with student feedback are adjusted to continuously optimize the accuracy and adaptability of the recommendation system. At the same time, based on students’ feedback information, the matching threshold is dynamically adjusted to improve the accuracy of recommendations and students’ satisfaction.

7. The educational course recommendation system according to claim 1, characterized in that: The system is equipped with a personalized recommendation mode switching function. Students can switch between the precise recommendation mode and the exploration recommendation mode according to their own needs. In the precise recommendation mode, the recommendation algorithm focuses more on matching students based on their existing characteristics and historical data to provide courses that are highly consistent with students' current interests and abilities. In the exploration recommendation mode, the system will appropriately lower the matching threshold and introduce some novel and challenging courses to encourage students to expand their learning areas and explore unknown knowledge. At the same time, the novelty and challenge indicators of the courses are clearly marked in the recommendation list to help students better understand the characteristics of the courses. The novelty index is determined by calculating the similarity between the course content and the courses that the students have studied. Let the novelty index be N, and the calculation formula is: where u it is the relevant feature value of the student on the course feature t. The challenge index is evaluated according to the difficulty level of the course and the gap in the student’s cognitive level. Let the challenge index be C h , the calculation formula is Among them C ci is the student’s current cognitive level assessment value, C ti is the difficulty level evaluation value of the course, C max and C min They are the maximum and minimum difficulty levels of the courses in the system respectively.

8. The educational course recommendation system according to claim 1, characterized in that: The course feature extraction module and the student feature analysis module adopt a distributed computing architecture. The specific distributed computing strategy is: the course data and student data are divided into multiple computing nodes according to the data volume and relevance. Each computing node processes the feature extraction and analysis tasks of part of the data in parallel, and then summarizes the results to the main node for integration. When dividing the data, a hash algorithm is used to ensure the balanced distribution of the data. At the same time, considering the relevance of the data, related data is distributed to the same computing node or adjacent computing nodes to reduce data transmission and communication overhead and improve the operating efficiency and response speed of the entire system.

9. The educational course recommendation system according to claim 1, characterized in that: The system has an interface with the school's teaching management system, which can obtain information about students' course selection, teaching schedule, and the school's education and teaching plan in real time, and integrate this information into the recommendation algorithm to make the recommended courses consistent with the school's overall teaching plan and students' actual learning progress. At the same time, according to the school's teaching focus and training goals, the weight parameters in the recommendation algorithm are adjusted to highlight the course recommendations that are in line with the school's education direction. The weight adjustment coefficient of the school's key courses is ω s , when the course belongs to the educational direction that the school focuses on, w f t=w ft ×(1+ω s ), where w f t is the weight of course feature t.

10. The educational course recommendation system according to claim 1, characterized in that: The system is equipped with a security and privacy protection module, which uses encryption technology to encrypt, store and transmit students' learning data and personal information to ensure the security and confidentiality of the data. During the use of data, it follows a strict privacy policy, anonymizes the data, removes sensitive information that can directly identify the student's identity, and only retains the necessary feature information for the calculation of the recommendation algorithm. At the same time, a data access rights management system is established to restrict unauthorized personnel from accessing and using student data, protect students' privacy rights and interests from infringement, and use blockchain technology to store and trace data operation records to ensure the integrity and auditability of the data and prevent data from being tampered with or abused.

Images