Intelligent campus management system for comprehensive quality cultivation of students

Abstract

The invention relates to the technical field of smart campus management, in particular to a smart campus management system oriented to student comprehensive quality cultivation, which comprises the steps of: performing analysis processing of each preset comprehensive quality index on multi-modal data of a student through an optimized fuzzy comprehensive evaluation method to obtain a comprehensive quality evaluation result of the student; when the evaluation grade of the quality type in the comprehensive quality evaluation result belongs to a preset quality lack grade; and according to the student information and the lacked quality type, recommending a plurality of customized resource information meeting the student cultivation demand for the student from the quality cultivation resource information, through the above mode, the method solves the problem that a traditional college student comprehensive quality evaluation method generally depends on a subjective questionnaire or a single data source, so that the quality of the college student is influenced. The problems that the comprehensive quality of the college students is rarely evaluated in detail from multiple aspects, objectivity and comprehensiveness are lacked, multi-dimensional features of the college students cannot be truly reflected, and the comprehensive quality lacked by the students cannot be cultivated in the prior art are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of smart campus management, and in particular to a smart campus management system for cultivating students' comprehensive qualities. Background Art

[0002] With the widespread and growing popularity of higher education, the comprehensive quality of college students has become a key indicator in educational evaluation. This comprehensive quality primarily encompasses professional competence, physical health, social skills, practical skills, and psychological well-being. Physical health emphasizes physical function and health management. Regular exercise, a balanced diet, and good sleep and rest habits enhance endurance and stress tolerance, providing the physiological support for long-term learning and work. Practical skills encompass practical skills and professionalism, such as engaging in social practice, internships, or volunteering. This fosters a spirit of hard work, an understanding of the value of labor, and enhanced teamwork and problem-solving abilities. Professional competence refers to a student's in-depth understanding of their discipline, including technical application skills, industry knowledge, and career development plans, such as obtaining professional qualifications and participating in industry competitions. Social skills refer to the ability to effectively communicate, express themselves, and self-regulate in interpersonal interactions, using language, emotions, and body language. These skills are crucial for adapting to society and achieving personal goals. Psychological quality refers to the comprehensive manifestation of psychological abilities and personality traits formed through acquired environment, education and practical training. It is the core factor affecting individual behavior, mental health and social adaptation. It mainly focuses on emotional management ability, flexibility and tolerance in coping with changes, stress, interpersonal relationships, and self-awareness.

[0003] However, traditional methods for evaluating the comprehensive qualities of college students usually rely only on subjective questionnaires or a single data source, and rarely conduct detailed evaluations of the comprehensive qualities of college students from multiple aspects. They lack objectivity and comprehensiveness, cannot truly reflect the multi-dimensional characteristics of college students, and thus cannot address the problem of cultivating students' lack of comprehensive qualities.

[0004] Therefore, the present invention provides a smart campus management system for cultivating students' comprehensive qualities to solve the above problems. Summary of the Invention

[0005] In view of the above situation, in order to overcome the defects of the existing technology, the present invention provides a smart campus management system for cultivating students' comprehensive qualities, so as to solve the problem that the traditional comprehensive quality evaluation method of college students usually only relies on subjective questionnaires or a single data source, rarely conducts detailed evaluation of the comprehensive qualities of college students from multiple aspects, lacks objectivity and comprehensiveness, cannot truly reflect the multi-dimensional characteristics of college students, and thus cannot cultivate the comprehensive qualities that students lack.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is:

[0007] A smart campus management system for cultivating students' comprehensive qualities, including:

[0008] A data acquisition module, which acquires students' multimodal data, including academic data, health data, behavioral data, and social data;

[0009] The quality assessment module uses an improved fuzzy comprehensive evaluation method to analyze and process the multimodal data based on various preset indicators to obtain the comprehensive quality assessment results of students;

[0010] The quality training module obtains the quality type whose evaluation level in the comprehensive quality assessment results belongs to the preset quality deficiency level; determines the quality training resource information that can be participated in based on the quality type and student information; and selects multiple customized resource information whose matching degree meets the preset conditions from the quality training resource information based on the comprehensive quality assessment results and student information.

[0011] Preferably, the improved fuzzy comprehensive evaluation method is used to analyze and process the multimodal data for each preset indicator to obtain the student's comprehensive quality assessment result, including: the preset indicators include academic ability indicators, physical health indicators, social ability indicators, practical ability indicators and psychological quality indicators; the multimodal data is classified and processed according to the demand data information of each preset indicator to obtain the factor data of each preset indicator; the factor data is processed and analyzed according to the preset processing rules of each indicator to determine the weight vector; and the corresponding fuzzy relationship matrix is ​​determined according to the membership function of each preset indicator; the γ operator is used to perform fuzzy operation on the weight vector and the fuzzy relationship matrix to obtain a comprehensive evaluation vector; and the comprehensive quality assessment result is determined according to the comprehensive evaluation vector.

[0012] Preferably, the multimodal data is classified and processed according to the demand data information of each preset indicator to obtain the factor data of each preset indicator, including: determining the associated data of each preset indicator in the multimodal data according to the preset indicator map; according to each preset indicator, using a video analysis model based on the attention mechanism to identify student behavior in the video data in the multimodal data to obtain feature data; preprocessing the non-video data and feature data in the multimodal data to obtain the factor data of each preset indicator.

[0013] Preferably, the factor data is processed and analyzed according to the preset processing rules of each indicator to determine the weight vector, including: using the entropy weight method to determine the entropy value of the factor data, and determining the entropy weight based on the entropy value to obtain the initial weight; according to the relationship between each indicator and the time period, the initial weight is adjusted using the time attenuation factor to obtain multiple single indicator weights; and the single indicator weights are integrated to obtain the weight vector.

[0014] Preferably, the method of determining the corresponding fuzzy relationship matrix based on the membership function of each preset indicator includes: determining the academic membership of the academic ability indicator using the Gaussian kernel density estimation formula; and determining the membership of the psychological quality indicator, the social ability indicator and the practical ability indicator using the Logistic function; determining the physical membership according to the preset standards of the factor data of the physical health indicator; and constructing a fuzzy relationship matrix based on each membership.

[0015] Preferably, the γ operator is used to perform fuzzy operation on the weight vector and the fuzzy relationship matrix to obtain the calculation formula of the comprehensive evaluation vector:

[0016]

[0017] Among them, B k is the final membership degree corresponding to the kth evaluation level, ω j is the weight of the jth preset indicator, r jk is the membership of the jth preset indicator to the kth evaluation level, μ is the balance parameter in the γ operator, and m is the total number of preset indicators.

[0018] Preferably, the determining of the quality training resource information available based on the quality type and student information includes: determining available query features based on the comprehensive quality assessment results and student information; and determining the quality training resource information from the quality training resource pool based on the query features.

[0019] Preferably, the method of selecting multiple customized resource information whose matching degree meets preset conditions from the quality training resource information based on the comprehensive quality assessment results and student information includes: determining the first-level customized resource information based on the semantic similarity between the quality type and the resource type; determining the second-level customized resource information from the first-level customized resource information based on the matching degree between the student ability corresponding to the student information and the resource difficulty level; predicting interest preferences through matrix decomposition based on the student's historical behavior data and questionnaire data; and determining multiple customized resource information from the second-level customized resource information based on the interest preferences.

[0020] Preferably, the second-level customized resource information is determined from the first-level customized resource information based on the degree of matching between the student ability corresponding to the student information and the resource difficulty level, including: using weighted cosine similarity to process the student ability vector and the resource difficulty vector to determine the matching degree; selecting a set number of resource information with the highest matching degree from the first-level customized resource information as the second-level customized resource information.

[0021] Preferably, the method predicts interest preferences through matrix decomposition based on students' historical behavior data and questionnaire data, including: constructing a historical behavior matrix based on students' historical participation frequency in activities based on historical behavior data; constructing a questionnaire feature matrix based on students' explicit ratings of questionnaire features based on questionnaire data; mapping the historical behavior matrix and the questionnaire feature matrix to the same low-dimensional latent space through matrix decomposition to obtain students' potential interest matrix, activity potential feature matrix and questionnaire feature latent factor matrix; iteratively optimizing the students' potential interest matrix, activity potential feature matrix and questionnaire feature latent factor matrix, and determining a low-rank approximate matrix when the loss function change is less than a preset threshold or reaches the maximum number of iterations; and determining students' interest preference for activities based on the low-rank approximate matrix.

[0022] The beneficial effects of the present invention are:

[0023] 1. The present invention uses an optimized fuzzy comprehensive evaluation method to analyze and process the multimodal data of students during their school years for each preset comprehensive quality index to obtain the student's comprehensive quality evaluation result; when the comprehensive quality evaluation result contains one or more quality types whose evaluation level belongs to the preset quality deficiency level; based on the student information and the type of quality that is lacking, the present invention recommends multiple customized resource information that meets the student's training needs from the quality training resource information, so that the student can select the corresponding quality training resources from them to train the quality that is lacking, and comprehensively train the student's comprehensive quality. Through the above method, the present invention solves the problem that traditional comprehensive quality evaluation methods for college students usually rely only on subjective questionnaires or a single data source, rarely conduct detailed evaluations of the comprehensive quality of college students from multiple aspects, lack objectivity and comprehensiveness, cannot truly reflect the multidimensional characteristics of college students, and thus cannot train students for the comprehensive qualities that they lack.

[0024] 2. When recommending customized resource information that meets students' needs, the present invention adopts a multi-layer screening method to select multiple customized resource information that meets the students' current professional level, learning ability and interest preferences from the quality training resource information, so that students can choose the courses or activities they want to take from these customized resource information, so as to cultivate the quality types that students lack, and cultivate students into comprehensive talents with all quality types meeting the standards, which helps to improve students' social adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 This is a schematic diagram of a smart campus management system for cultivating students' comprehensive qualities according to the present invention. DETAILED DESCRIPTION

[0026] The following will refer to the attached Figure 1The embodiments of the present invention are described in detail. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0027] A smart campus management system for cultivating students' comprehensive qualities, as shown in the attached Figure 1 Shown, including:

[0028] The data acquisition module acquires students' multimodal data, which includes academic data, health data, behavioral data, and social data. Among them, the academic data comes from the academic affairs system, including grades, homework, and classroom interactions; the health data comes from physical examination reports, physical test reports, etc., including height, weight, heart rate, lung capacity, etc.; behavioral data includes campus card record data and questionnaire survey data. Campus card record data includes library visit data, club activity data, etc.; social data includes roommate evaluations, teacher-student evaluations, forum interaction data, etc.

[0029] Among them, multimodal data related to academic ability include course grades, homework completion rate, frequency of classroom interaction, knowledge graph correlation, that is, one or more of interdisciplinary performance; multimodal data related to physical health quality include physical test score reports, posture recognition results in sports videos; multimodal data related to practical ability include course design scores, corporate internship evaluations, simulation experiment data (such as programming accuracy, engineering drawing standardization), etc.; multimodal data related to psychological quality include student peer evaluation texts (sentiment analysis), stress test scores, campus card consumption records (integrity), students' online comments, etc.

[0030] The quality assessment module uses an improved fuzzy comprehensive evaluation method to analyze and process the multimodal data based on various preset indicators to obtain the comprehensive quality assessment results of students;

[0031] The quality training module obtains the quality type whose evaluation level in the comprehensive quality assessment results belongs to the preset quality deficiency level; determines the quality training resource information that can be participated in based on the quality type and student information; selects multiple customized resource information whose matching degree meets the preset conditions from the quality training resource information based on the comprehensive quality assessment results and student information; and recommends the customized resource information to students so that they can choose the corresponding quality training resources from it to carry out training in the quality areas they lack.

[0032] The present invention uses an optimized fuzzy comprehensive evaluation method to analyze and process the multimodal data of students during their school years for each preset comprehensive quality index to obtain the student's comprehensive quality evaluation result; when the comprehensive quality evaluation result contains one or more quality types whose evaluation level belongs to the preset quality deficiency level; based on the student information and the type of quality that is lacking, the present invention recommends multiple customized resource information that meets the student's training needs from the quality training resource information, so that the student can select the corresponding quality training resources from them to train the quality that is lacking, and comprehensively train the student's comprehensive quality. Through the above method, the present invention solves the problem that traditional comprehensive quality evaluation methods for college students usually rely only on subjective questionnaires or a single data source, rarely conduct detailed evaluation of the comprehensive quality of college students from multiple aspects, lack objectivity and comprehensiveness, cannot truly reflect the multidimensional characteristics of college students, and thus cannot train students for the comprehensive qualities that they lack.

[0033] Furthermore, when recommending customized resource information that meets students' needs to students, the present invention adopts a multi-layer screening method to select multiple customized resource information that meets the students' current professional level, learning ability and interest preferences from the quality training resource information, so that students can choose the courses or activities they want to take from these customized resource information, so as to cultivate the quality types that students lack, and cultivate students into comprehensive talents with all quality types meeting the standards, which helps to improve students' social adaptability.

[0034] In one embodiment of the present invention, an improved fuzzy comprehensive evaluation method is used to analyze and process multimodal data for each preset indicator to obtain a comprehensive quality assessment result of the student, including: the preset indicators include academic ability indicators, physical health indicators, social ability indicators, practical ability indicators and psychological quality indicators; the multimodal data is classified and processed according to the demand data information of each preset indicator to obtain factor data of each preset indicator; the factor data is processed and analyzed according to the preset processing rules of each indicator to determine a weight vector; and the corresponding fuzzy relationship matrix is ​​determined according to the membership function of each preset indicator; the gamma operator is used to perform fuzzy operation on the weight vector and the fuzzy relationship matrix to obtain a comprehensive evaluation vector; and the comprehensive quality assessment result is determined based on the comprehensive evaluation vector.

[0035] Through the setting method of this embodiment, the present invention can use the improved fuzzy comprehensive evaluation method to analyze and process the multimodal data for various preset indicators, and obtain the comprehensive quality evaluation results of the students in a comprehensive, reasonable and objective manner, thereby facilitating the subsequent recommendation of quality training resource information for the quality types that the students lack, which helps to improve the students' comprehensive quality capabilities.

[0036] Furthermore, the multimodal data is classified and processed according to the demand data information of each preset indicator to obtain the factor data of each preset indicator, including: determining the associated data of each preset indicator in the multimodal data according to the preset indicator map; according to each preset indicator, using a video analysis model based on the attention mechanism to identify student behavior in the video data in the multimodal data to obtain feature data; preprocessing the non-video data and feature data in the multimodal data to obtain the factor data of each preset indicator.

[0037] The preset indicator map includes knowledge maps of different indicators and common courses, activities, or practical projects. The factor data in the knowledge map is one or more courses in a common course, an action or operation in a certain time period in an activity, and one or more practical sub-projects in a practical project. After determining the associated data of each preset indicator, a video analysis model based on an attention mechanism is used to identify student behavior in the video data of the multimodal data according to each preset indicator. The following includes: in terms of video recognition and analysis of academic ability, the frequency of students' interactions in class, the accuracy or relevance of questions during interaction, etc., to obtain characteristic data on academic ability; in terms of physical health quality, based on the video data of students in sports, the accuracy of students' postures and movements, the number of training sessions, the duration, etc. are identified through a video analysis model focusing on movement and posture recognition to obtain characteristic data on physical health quality; non-video data includes questionnaire data corresponding to each preset indicator, grade data, peer review text (sentiment analysis), stress test scores, campus card consumption records (integrity), physical test score reports, homework completion rate, frequency of classroom interaction, and students' online comments. During preprocessing, in addition to classifying the data according to various preset indicators, different types of non-video data and feature data are formatted uniformly and quantified to ultimately obtain quantified factor data.

[0038] Through the setting method of this embodiment, the present invention can objectively perform targeted analysis and processing on the students' multimodal data to obtain factor data corresponding to different preset indicators, so as to subsequently use the improved fuzzy comprehensive evaluation method to analyze and process the factor data and obtain comprehensive and objective analysis results of the students' comprehensive qualities.

[0039] In one embodiment of the present invention, the factor data is processed and analyzed according to the preset processing rules of each indicator to determine the weight vector, including: using the entropy weight method to determine the entropy value of the factor data, and determining the entropy weight based on the entropy value to obtain the initial weight; according to the relationship between each indicator and the time period, the initial weight is adjusted using the time attenuation factor to obtain multiple single indicator weights; and the single indicator weights are integrated to obtain the weight vector.

[0040] In this embodiment, the i-th entropy value Ei The calculation formula is:

[0041] Among them, m is the total number of factor data of the i-th preset indicator, p i is the standardized value ratio of the factor data of the i-th preset indicator;

[0042] The i-th entropy weight W i The calculation formula is:

[0043] Among them, n is the total number of entropy values, that is, the total number of preset indicators, E j is the jth entropy value;

[0044] The formula for calculating the weight of a single indicator using the time decay factor at the beginning of the semester is:

[0045] W(t)=W0·e -κt ,

[0046] The formula for calculating the weight of a single indicator using the time decay factor at the end of the semester is:

[0047]

[0048] Among them, κ is the attenuation coefficient, T is the total duration, t is the time variable of the current semester, and W0 is the initial weight.

[0049] Among them, the first preset weight is the initial weight of the preset indicator determined by the entire process of multiple experts analyzing factor data and comprehensive evaluation; since the initial weight is a fixed value, the present invention adopts the entropy weight method and time decay factor to adjust the initial weight according to the factor data of the preset indicator and the current learning time period, so that the final determined single indicator weight is more in line with the current actual situation, thereby improving the accuracy of the weight value.

[0050] Through the setting method of this embodiment, the present invention can avoid adjusting the weight value in the fuzzy comprehensive evaluation method by combining the entropy weight method and the time attenuation factor, and can adapt to the needs of dynamic changes in students' qualities, thereby helping to improve the accuracy of the final comprehensive quality assessment results.

[0051] Furthermore, the method of determining the corresponding fuzzy relationship matrix based on the membership function of each preset indicator includes: using the Gaussian kernel density estimation formula to determine the academic membership of the academic ability indicator; and using the Logistic function to determine the membership of the psychological quality indicator, the social ability indicator and the practical ability indicator; determining the physical membership according to the preset standards of the factor data of the physical health indicator; and constructing a fuzzy relationship matrix based on each membership.

[0052] In this embodiment, the formula for calculating the membership degree ψ1(x) by Gaussian kernel density estimation is:

[0053]

[0054] Among them, ψ is the mean of the student scores of the corresponding subject, σ is the variance of the student scores of the corresponding subject, and x is the student scores of the corresponding subject.

[0055] The formula for calculating the membership degree ψ2(s) using the Logistic function is:

[0056]

[0057] Among them, s0 is the average evaluation value of the evaluation item, k c is the weight coefficient, which can be adjusted according to actual conditions, and s is the evaluation score of the student evaluation item.

[0058] Because academic performance has a linear relationship, a Gaussian kernel density estimation formula was used to determine the academic membership of the academic ability indicators, determine the student's membership in terms of academic performance, and determine the evaluation grade of the academic ability indicators. Because the membership and factor data of the psychological quality indicators, social skills indicators, and practical ability indicators have a nonlinear relationship, the logistic function was used to objectively determine the membership of each indicator, thereby reasonably determining the evaluation grade corresponding to each individual indicator. The physical membership was determined based on the preset standards of the factor data of the physical health indicators. The preset standards refer to the standards for blood pressure assessment, heart rate assessment, physical health index assessment, and vital capacity assessment. The quantitative data obtained from the video was analyzed using the logistic function to obtain the corresponding membership. For the membership of the same preset indicator, the preset weights and membership were averaged using the weighted average method to obtain the average membership of the indicator. A fuzzy relationship matrix was then constructed based on the membership.

[0059] Through the above implementation, the present invention can set a targeted membership function for different preset indicators, which helps to improve the accuracy of the membership of each preset indicator, and further helps to improve the accuracy of the final comprehensive quality assessment result.

[0060] Furthermore, the γ operator is used to perform fuzzy operations on the weight vector and the fuzzy relationship matrix, and the calculation formula for the comprehensive evaluation vector is obtained as follows:

[0061]

[0062] Among them, B k is the final membership degree corresponding to the kth evaluation level, ω j is the weight of the jth preset indicator, r jkis the membership of the jth preset indicator to the kth evaluation level, μ is the balance parameter in the γ operator, m is the total number of preset indicators, and the total number of evaluation levels is n.

[0063] In this embodiment, by The γ operator preserves the correlation between the main factors. For example, when all pre-defined indicators are poor, the product term approaches 0, avoiding the information loss of the traditional weighted summation method. The γ value can be adjusted according to the scenario. For example, during the final evaluation, μ can be increased to strengthen the influence of key pre-defined indicators (such as test scores). Moreover, compared with a single Max-Min operator, the γ operator takes into account both the weighted sum and the product term, reducing the interference of extreme values ​​on the results. In short, it preserves the key information of the main factors while avoiding over-reliance on a single indicator.

[0064] Through the above-mentioned implementation, the present invention can analyze and process the multimodal data of students during their school years for each preset comprehensive quality index through an optimized fuzzy comprehensive evaluation method to obtain the student's comprehensive quality assessment result; then, when the comprehensive quality assessment result contains one or more quality types whose assessment level belongs to the preset quality deficiency level; based on the student information and the type of quality that is lacking, the present invention recommends multiple customized resource information that meets the student's training needs from the quality training resource information, so that the student can select the corresponding quality training resources from them to train the quality deficiencies and comprehensively cultivate the student's comprehensive quality. The present invention solves the problem that traditional comprehensive quality evaluation methods for college students usually rely only on subjective questionnaires or a single data source, rarely conduct detailed evaluations of the comprehensive quality of college students from multiple aspects, lack objectivity and comprehensiveness, and cannot truly reflect the multidimensional characteristics of college students.

[0065] In one embodiment of the present invention, the quality training resource information available for participation is determined based on the quality type and student information, including: determining available query features based on the comprehensive quality assessment results and student information; and determining the quality training resource information from the quality training resource pool based on the query features.

[0066] The query features include the quality type that students need to improve in the comprehensive quality assessment results, the student's major and grade, gender, college name, etc. The quality training resource pool is a database record of the school's existing educational resources, which includes quality type tags.

[0067] In one embodiment of the present invention, multiple customized resource information with matching degrees that meet preset conditions are selected from quality training resource information based on comprehensive quality assessment results and student information, including: determining first-level customized resource information based on semantic similarity between quality type and resource type; determining second-level customized resource information from the first-level customized resource information based on the matching degree between student ability corresponding to student information and resource difficulty level; predicting interest preferences through matrix decomposition based on students' historical behavior data and questionnaire data; and determining multiple customized resource information from the second-level customized resource information based on interest preferences.

[0068] Among them, the specific process of determining the first-level customized resource information based on the semantic similarity between the quality type and the resource type is: using the association analysis method to perform association analysis on the quality type and the type label in the resource information, determine the semantic similarity, and use the resource information with a similarity greater than 70% in the quality training resource information as the first-level customized resource information for preliminary screening.

[0069] Furthermore, based on the degree of matching between the student abilities and resource difficulty levels corresponding to the student information, the second-level customized resource information is determined from the first-level customized resource information, including: using weighted cosine similarity to process the student ability vector and the resource difficulty vector to determine the matching degree; and selecting a set number of resource information with the highest matching degree from the first-level customized resource information as the second-level customized resource information.

[0070] The student information-based student ability assessment method involves quantifying the student's historical data to obtain scores for each ability. The scores for each ability form a student ability vector. The resource difficulty vector is formed based on the difficulty level of the first-level customized resource information.

[0071] Through the above-mentioned method, the present invention can determine the second-level customized resource information from the first-level customized resource information by using weighted cosine similarity, and quickly screen out the second-level customized resource information required by students.

[0072] Furthermore, based on the students' historical behavior data and questionnaire data, interest preferences are predicted through matrix decomposition, including: constructing a historical behavior matrix based on the students' historical participation frequency in activities according to the historical behavior data; constructing a questionnaire feature matrix based on the students' explicit ratings of questionnaire features according to the questionnaire data; mapping the historical behavior matrix and the questionnaire feature matrix to the same low-dimensional latent space through matrix decomposition to obtain the students' potential interest matrix, activity potential feature matrix and questionnaire feature latent factor matrix; iteratively optimizing the students' potential interest matrix, activity potential feature matrix and questionnaire feature latent factor matrix, and determining a low-rank approximate matrix when the loss function change is less than a preset threshold or reaches the maximum number of iterations; and determining the students' interest preference for activities based on the low-rank approximate matrix.

[0073] The preset threshold or the maximum number of iterations may be adjusted according to actual conditions.

[0074] Specifically, the loss function The calculation formula is:

[0075]

[0076] Among them, R is the historical behavior matrix, U is the student potential interest matrix, V is the activity potential feature matrix, α is the weight for balancing historical behavior and questionnaire data, Q is the questionnaire feature matrix, W is the questionnaire feature latent factor matrix, β is the regularization coefficient to prevent overfitting, F is the latent factor dimension, ||·|| F Represents the Frobenius norm of the matrix. Regularization terms are used to avoid excessive complexity of the latent factor matrix.

[0077] Through matrix decomposition, historical behavior and questionnaire data are mapped to the same low-dimensional latent space. During the reconstruction of historical behavior, UV T Approximate R, during the reconstruction of questionnaire features, UW T Approaching Q.

[0078] Then, by alternately fixing the variables and solving the closed-form solution, U, V, and W are gradually optimized, where the V minimization objective is:

[0079] Taking the derivative of V and setting it to 0, we get the update formula of the activity feature matrix V:

[0080] V=(U T U+βI) -1 U T R,

[0081] Where I is the intermediate matrix generated by the derivation.

[0082] The W minimization objective is:

[0083] Taking the derivative of W and setting it to 0, we get the update formula of the questionnaire feature matrix W:

[0084] W=(αU T U+βI) -1 αU T Q,

[0085] The U minimization objective is:

[0086] Taking the derivative of U and setting it to 0, we get the update formula of the student's potential interest matrix U:

[0087] U=(VVT +αWW T +βI) -1 (VR T +αWQ T ),

[0088] The final student interest preference is approximated by low rank get:

[0089] in, represents the predicted interest preference of student i for activity j.

[0090] When the loss function change is less than the threshold or the maximum number of iterations is reached, the process stops and the final low-rank approximation matrix is ​​obtained; then, the students' interest preference for activities is generated based on the low-rank approximation matrix.

[0091] Through the setting method of this embodiment, when recommending customized resource information that meets the needs of students, the present invention adopts a multi-layer screening method to select multiple customized resource information that meets the students' current professional level, learning ability and interest preferences from the quality training resource information, so that students can choose the courses or activities they want to take from these customized resource information, so as to cultivate the quality types that students lack, and cultivate students into comprehensive talents with all quality types meeting the standards, which helps to improve students' social adaptability.

[0092] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system comprising at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0093] It should be noted that, in the description of the present invention, the terms "first", "second" and "third" are used for descriptive purposes only and should not be understood as indicating or implying relative importance.

[0094] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0095] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0096] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0097] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0098] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact through a communication network. The client-server relationship arises through computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, a server in a distributed system, or a server integrated with a blockchain.

[0099] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.

Claims

1. A smart campus management system for cultivating students' comprehensive qualities, characterized by: include: A data acquisition module, which acquires students' multimodal data, including academic data, health data, behavioral data, and social data; The quality assessment module uses an improved fuzzy comprehensive evaluation method to analyze and process the multimodal data based on various preset indicators to obtain the comprehensive quality assessment results of students; The quality training module obtains the quality type whose evaluation level in the comprehensive quality assessment results belongs to the preset quality deficiency level; determines the quality training resource information that can be participated in based on the quality type and student information; and selects multiple customized resource information whose matching degree meets the preset conditions from the quality training resource information based on the comprehensive quality assessment results and student information.

2. The smart campus management system according to claim 1, characterized in that: The improved fuzzy comprehensive evaluation method is used to analyze and process multimodal data for each preset indicator to obtain a comprehensive quality assessment result of the student, including: the preset indicators include academic ability indicators, physical health indicators, social ability indicators, practical ability indicators and psychological quality indicators; the multimodal data is classified and processed according to the required data information of each preset indicator to obtain factor data of each preset indicator; the factor data is processed and analyzed according to the preset processing rules of each indicator to determine a weight vector; and the corresponding fuzzy relationship matrix is ​​determined according to the membership function of each preset indicator; the gamma operator is used to perform fuzzy operation on the weight vector and the fuzzy relationship matrix to obtain a comprehensive evaluation vector; and the comprehensive quality assessment result is determined based on the comprehensive evaluation vector.

3. The smart campus management system according to claim 2, characterized in that: The method of classifying and processing the multimodal data according to the demand data information of each preset indicator to obtain the factor data of each preset indicator includes: determining the associated data of each preset indicator in the multimodal data according to the preset indicator map; according to each preset indicator, using a video analysis model based on the attention mechanism to identify student behavior in the video data in the multimodal data to obtain feature data; preprocessing the non-video data and feature data in the multimodal data to obtain the factor data of each preset indicator.

4. The smart campus management system according to claim 2, characterized in that: The method of processing and analyzing the factor data according to the preset processing rules of each indicator to determine the weight vector includes: using the entropy weight method to determine the entropy value of the factor data, and determining the entropy weight based on the entropy value to obtain the initial weight; according to the relationship between each indicator and the time period, using the time decay factor to adjust the initial weight to obtain multiple single indicator weights; integrating the single indicator weights to obtain the weight vector.

5. The smart campus management system according to claim 2, characterized in that: The method of determining the corresponding fuzzy relationship matrix based on the membership function of each preset indicator includes: using the Gaussian kernel density estimation formula to determine the academic membership of the academic ability indicator; and using the logistic function to determine the membership of the psychological quality indicator, the social ability indicator and the practical ability indicator; determining the physical membership according to the preset standard of the factor data of the physical health indicator; and constructing the fuzzy relationship matrix according to each membership.

6. The smart campus management system according to claim 2, characterized in that: The γ operator is used to perform fuzzy operation on the weight vector and the fuzzy relationship matrix to obtain the calculation formula of the comprehensive evaluation vector: Among them, B k is the final membership degree corresponding to the kth evaluation level, ω j is the weight of the jth preset indicator, r jk is the membership of the jth preset indicator to the kth evaluation level, μ is the balance parameter in the γ operator, and m is the total number of preset indicators.

7. The smart campus management system according to claim 1, characterized in that: The method of determining the quality training resource information available based on the quality type and student information includes: determining available query features based on the comprehensive quality assessment results and student information; and determining the quality training resource information from the quality training resource pool based on the query features.

8. The smart campus management system according to claim 1, characterized in that: The method of selecting multiple customized resource information whose matching degree meets preset conditions from the quality training resource information based on the comprehensive quality assessment results and student information includes: determining the first-level customized resource information based on the semantic similarity between the quality type and the resource type; determining the second-level customized resource information from the first-level customized resource information based on the matching degree between the student ability corresponding to the student information and the resource difficulty level; predicting interest preferences through matrix decomposition based on the student's historical behavior data and questionnaire data; and determining multiple customized resource information from the second-level customized resource information based on the interest preferences.

9. The smart campus management system according to claim 8, characterized in that: The method of determining the second-level customized resource information from the first-level customized resource information based on the degree of matching between the student ability and the resource difficulty level corresponding to the student information includes: using weighted cosine similarity to process the student ability vector and the resource difficulty vector to determine the matching degree; and selecting a set number of resource information with the highest matching degree from the first-level customized resource information as the second-level customized resource information.

10. The smart campus management system according to claim 8, characterized in that: The method predicts interest preferences through matrix decomposition based on students' historical behavior data and questionnaire data, including: constructing a historical behavior matrix based on students' historical participation frequency in activities according to the historical behavior data; constructing a questionnaire feature matrix based on students' explicit ratings of questionnaire features according to the questionnaire data; mapping the historical behavior matrix and the questionnaire feature matrix to the same low-dimensional latent space through matrix decomposition to obtain students' potential interest matrix, activity potential feature matrix and questionnaire feature latent factor matrix; iteratively optimizing the students' potential interest matrix, activity potential feature matrix and questionnaire feature latent factor matrix, and determining a low-rank approximate matrix when the loss function change is less than a preset threshold or reaches the maximum number of iterations; and determining students' interest preference for activities based on the low-rank approximate matrix.

Images