Intelligent management system based on data analysis

Abstract

The invention discloses an intelligent management system based on data analysis, and belongs to the field of learning management systems, and the system comprises a weak subject analysis module which is used for extracting each subject score of a target student from a historical examination score database; the assistance matching module is in communication connection with the weak subject analysis module and is used for establishing a voluntary application one-to-one assistance relationship for the students with the weak subjects; the wrong question path generation module is used for analyzing the examination wrong questions of the target student in recent three months, counting knowledge point weights and constructing a dynamic learning path; the risk early warning module is used for detecting students whose scores decline in a cliff mode and starting three-level risk early warning through behavior and state analysis of campus monitoring; and the notification execution engine is used for triggering teacher appointment conversation, parent communication, seat adjustment and learning resource pushing operation according to the analysis result. According to the invention, quantitative evaluation can be carried out on dynamic change of subject strength, and then intervention necessity is automatically judged.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of learning management systems, in particular to an intelligent management system based on data analysis. BACKGROUND

[0002] The intelligent learning management system refers to a comprehensive management platform for dynamically monitoring, diagnosing problems and scheduling resources for students' learning process through data collection, machine learning and behavior analysis technology. The core goal is to achieve personalized education intervention through data-driven.

[0003] The existing patent 201910929592.8, a kind of individualized learning and intelligent training management system and method, including student end, teacher end, management end and background, student end is used to view student personal information, recommend learning resources to student, and examine the knowledge points learned;Teacher end is used to answer the questions raised by students, select course resources or provide relevant information of training tasks;Management end is used for training task management, resource audit and feedback analysis;Background is used for information collection, data processing and analysis, resource storage, reading and updating, student end includes registration login module, user information module, learning and training module, ability test module, teacher application module and information feedback module, teacher end includes knowledge answering module, training task module, test question resource upload module, learning resource upload module and integral module.

[0004] The existing technology has the following problems: the traditional system only counts scores and rankings, lacks a quantitative evaluation mechanism for the dynamic changes of subject strengths and weaknesses, and cannot automatically determine the necessity of intervention. Therefore, the present application provides an intelligent management system based on data analysis to solve the problems raised in the above background technology. SUMMARY

[0005] The purpose of the present application is to provide an intelligent management system based on data analysis, which can quantitatively evaluate the dynamic changes of subject strengths and weaknesses, and automatically determine the necessity of intervention to solve the problems raised in the above background technology.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical solutions: An intelligent management system based on data analysis, comprising: A weak subject analysis module for extracting the subject scores of a target student from a historical test score database, identifying the current weak subjects and strong subjects, and determining whether to take remedial measures based on the ranking changes of the subjects in previous tests; A help matching module in communication connection with the weak subject analysis module, for establishing a one-on-one help relationship for students with weak subjects through voluntary application, and matching eligible help objects through a class database; The wrong question path generation module is used for analyzing wrong questions of the target student in the last three months, counting knowledge point weights, and constructing a dynamic learning path. The risk early warning module is used for detecting students with a cliff-like decline in scores, and starting a three-level risk early warning through behavior and state analysis of campus monitoring. The notification execution engine is used for triggering teacher interviews, parent communication, seat adjustment and learning resource pushing operations according to the analysis results.

[0007] As a further scheme of the present application, the weak subject analysis module comprises: The remedy judgment unit: if the weak subject ranking does not reach the threshold value in the continuous two times of examination, it is determined that remedy is taken but the effect is insufficient; if the ranking continues to decline or remains unchanged, it is determined that remedy is not taken; The notification strategy unit: generating a teacher interview instruction when the effect is insufficient, and generating a class teacher home visit instruction when remedy is not taken; The weight calculation unit: calculating the score improvement weight W according to the formula k : ; Wherein, And is a preset adjustment coefficient, and , is the current ranking, is the previous ranking, N is the total number of students in the class, is the current score, is the previous score, and M is the full score of the subject; When W k <0.1, a teacher interview instruction is triggered.

[0008] As a further scheme of the present application, the help matching module performs the following operations: (a) receiving a student voluntary help application and archiving it to a class database; (b) when there is a target student candidate object in the database, matching according to four-dimensional rules: Rule 101: query the campus event record library, if there are records of fighting between the two parties, exclude; Rule 102: analyze the subject strength and weakness of the two parties, and the advantage subject of student A must contain the weak subject of student B, and the advantage subject of student B must contain the weak subject of student A; Rule 103: verify the gender through the student information library, and same-sex matching is preferred; Rule 104: compare the height value of the physical examination data, and exclude if the difference is more than 20 cm; (c) after successful matching, send a seat adjustment instruction to the class teacher.

[0009] As a further scheme of the present application: the wrong question path generation module comprises: A knowledge point weight calculation unit is configured to calculate a deduction weight according to a knowledge point involved in the wrong question, and a weight value calculation formula is: knowledge point deduction weight=(total deduction of the knowledge point / total deduction of all knowledge points)×100%; A path construction unit is configured to take the knowledge point with the maximum deduction weight as a starting point, and form a learning path by concatenating other knowledge points in descending order of a correlation degree between the knowledge points, wherein the correlation degree is derived from a textbook knowledge graph database; A resource pushing unit is configured to push course videos and exercise resources to a student terminal in a sequence along the learning path.

[0010] As a further scheme of the present application: further comprising a learning effect feedback module configured to perform the following operations after updating the learning path each time: (a) calculate a learning effect score of the student following the learning path in the previous stage based on the examination result data of this time, and the calculation method of the score is: subtract the score rate of the target knowledge point in the previous stage from the score rate of the target knowledge point in the current stage, and then divide by the full score value of the knowledge point and multiply by 100; wherein the score rate is defined as the ratio of the actual score of the knowledge point to the full score value of the knowledge point; (b) if the learning effect score is greater than or equal to 60 points, pack the learning path and the corresponding all learning resources of the previous stage as a successful case and store it to the cloud platform; (c) if the learning effect score is less than 60 points, execute the following remedial process: extract the weight values of all knowledge points in the current learning path of the target student to generate a feature sequence composed of the weight values; retrieve the 5 case paths with the highest similarity to the current feature sequence from the historical cases of the cloud platform; extract the learning effect scores recorded in the 5 case paths respectively; select the learning resource set associated with the case with the highest learning effect score from the above cases; push the learning resource set to the terminal of the target student for reference.

[0011] As a further scheme of the present application: the historical case retrieval process is executed according to the following steps: (i) extract all the knowledge points contained in the current learning path of the target student to form a current knowledge point set; (ii) retrieve the learning paths of all historical cases from the cloud platform, and extract the knowledge point set of each case path; (iii) calculate the coincidence degree between the current knowledge point set and the knowledge point set of each case: coincidence degree=(number of common knowledge points of the two sets) / (total amount of the current knowledge point set)×100%. (iv) Screen the top five case paths with the highest coincidence degree.

[0012] As a further scheme of the present application: the risk early warning module comprises: a behavior analysis unit that identifies the following abnormal states through campus monitoring video streams: (1) Abnormal mentality: showing a 80% alone time ratio for 3 consecutive days and a 50% drop in conversation frequency; (2) Campus bullying: showing physical conflict or clothing damage in the monitoring picture, or appearing with disheveled hair and dirty clothes after entering and leaving the unmonitored area; (3) Economic abnormality: showing a 5-day cafeteria consumption below the poverty line and a meal protein content of less than 30%; a warning response unit that sends differentiated warning information to the class teacher, the security department and the parents according to the type of abnormal state.

[0013] As a further scheme of the present application: the behavior analysis unit adopts multi-modal fusion recognition, specifically: an action capture subunit that extracts skeletal key points based on the OpenPose algorithm to identify crouching and pushing actions; an expression recognition subunit that analyzes facial features through a ResNet model to output a depression index; a speech analysis subunit that uses an LSTM network to detect voice tremor frequency and abnormal volume fluctuations.

[0014] As a further scheme of the present application: it further comprises a data security module, which comprises: a monitoring data desensitization unit for Gaussian blur processing of the face area in the monitoring picture; an operation audit unit for recording key operation logs through blockchain technology; a communication encryption unit for encrypting data transmission using the AES-256 algorithm.

[0015] As a further scheme of the present application: the notification execution engine contains an intelligent routing mechanism, specifically: teacher interview instructions are preferentially assigned to the student's free period on the timetable for the last 3 days; parent notification is preferentially notified by phone, and if the phone is not connected, a short message is sent; seat adjustment instructions are associated with a classroom three-dimensional model to automatically generate a seating arrangement that meets the height gradient distribution and does not block the blackboard.

[0016] Compared with the prior art, the present application has the following advantages: The application constructs an intelligent education management closed loop through multi-dimensional data analysis, significantly improving the accuracy and timeliness of teaching intervention. Among them, the weak subject precise diagnosis is realized based on the subject ranking dynamic threshold and the performance improvement weight formula (W k ) to automatically trigger differentiated intervention instructions (teacher interview / home visit). Through the four-dimensional support matching mechanism (historical relationship / subject complement / gender height / safety screening), the effectiveness of support is ensured and the seat intelligent adjustment is linked. In addition, the dynamic learning path is constructed relying on the wrong knowledge point weight, and the learning resource adaptive optimization is realized by combining the historical case similarity retrieval (coincidence algorithm). The three-level risk early warning is established by fusing multi-modal behavior recognition (skeletal action / expression / voice analysis), which prevents psychological crisis and campus bullying in real time. The intervention measures are automatically executed through the intelligent routing engine (schedule analysis / communication priority / three-dimensional ranking). In summary, the application can solve the problems of low efficiency of artificial analysis, lagging intervention, and inaccurate resource matching, improve academic performance, strengthen student safety protection, and ensure data compliance through monitoring desensitization and blockchain audit. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 It is a structural diagram of an intelligent management system based on data analysis. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0019] As mentioned in the background of the application, research has found that the existing system only counts scores and rankings, lacks quantitative evaluation mechanism for dynamic changes of subject strength and weakness, and cannot automatically determine the necessity of intervention, which has certain defects.

[0020] To solve the above-mentioned defects, the application discloses an intelligent management system based on data analysis, which can quantitatively evaluate the dynamic changes of subject strength and weakness, and then automatically determine the necessity of intervention.

[0021] The application will be described in detail below on how to solve the above-mentioned technical problems.

[0022] Please refer to Figure 1In the embodiment of the present application, an intelligent management system based on data analysis includes: a weak subject analysis module for extracting the subject student's performance in each subject from the historical test score database, identifying the current weak subject and the strong subject, and determining whether to take remedial measures based on the subject's previous test ranking changes; a help matching module in communication connection with the weak subject analysis module, for establishing a one-on-one help relationship for students with weak subjects by voluntarily applying through the class database to match the qualified help objects; a wrong question path generation module for analyzing the target student's wrong questions in the last three months, counting the knowledge point weight and constructing a dynamic learning path; a risk warning module for detecting students with a cliff-like drop in performance, starting a three-level risk warning through behavior and state analysis of campus monitoring; a notification execution engine for triggering teacher interview, parent communication, seat adjustment and learning resource pushing operation according to the analysis results. The present application constructs an intelligent management system including weak subject analysis, help matching, wrong question path generation, risk warning and notification execution engine, through the collaborative operation of multiple modules, realizes the accurate identification of students' subject short board, individualized help matching, dynamic learning path planning, behavior risk warning and automatic intervention execution, and comprehensively improves the teaching management efficiency and the student problem response speed.

[0023] In the present embodiment, the weak subject analysis module includes: a remediation judgment unit: if the weak subject ranking rises by less than the threshold value for two consecutive tests, it is determined that remediation is taken but the effect is insufficient; if the ranking continues to decline or remains unchanged, it is determined that remediation is not taken; a notification strategy unit: generating a teacher interview instruction when the effect is insufficient, and generating a class teacher home visit instruction when remediation is not taken; a weight calculation unit: calculating the performance improvement weight W according to the formula k : ; wherein, and is a preset adjustment coefficient, and , is the current ranking, is the previous ranking, N is the total number of students in the class, is the current score, is the previous score, and M is the subject full score; when W k <0.1, a teacher interview instruction is triggered. The setting quantifies the ranking change trend through the remediation judgment unit, generates differentiated intervention instructions through the notification strategy unit, and dynamically calculates the performance improvement weight W k through the weight calculation unit, realizes the objective evaluation of the weak subject remediation effect, and triggers the teacher interview based on the threshold value of W k <0.1, ensuring the scientificity and timeliness of the intervention measures.

[0024] In this embodiment, the assistance matching module performs the following operations: (a) Receives voluntary assistance applications from students and files them in the class database; (b) When target student candidates exist in the database, matches are performed according to four criteria: Rule 101: Query the campus event record database to exclude students with a history of fighting; Rule 102: Analyzes the strengths and weaknesses of both parties' academic disciplines, ensuring that both student A's strengths and weaknesses are met, and that both student B's strengths and weaknesses are met; Rule 103: Verifies gender through the student information database, prioritizing same-sex matches; Rule 104: Compares height values ​​from physical examination data; excludes students with a difference of more than 20 cm; (c) After a successful match, sends a seat adjustment instruction to the class teacher. This setup, based on a voluntary application mechanism, selects assistance candidates according to four criteria (no conflicting records, complementary academic disciplines, same-sex preference, and height compatibility), avoiding interpersonal conflicts and improving assistance efficiency. It also provides physical support for the assistance relationship by automatically triggering seat adjustment instructions.

[0025] In this embodiment, the wrong question path generation module includes: a knowledge point weight calculation unit, which is used to calculate the deduction weight based on the knowledge points involved in the wrong question. The weight value calculation formula is: knowledge point deduction weight = (total deduction for this knowledge point / total deduction for all knowledge points) × 100%; a path construction unit, which takes the knowledge point with the largest deduction weight as the starting point and connects other knowledge points from high to low according to the correlation between knowledge points to form a learning path. The correlation is derived from the textbook knowledge graph database; and a resource push unit, which pushes course videos and exercise resources to student terminals along the learning path. This setting quantifies the weak points of wrong questions through knowledge point weight calculation, constructs a dynamic learning path based on the correlation of the textbook knowledge graph, and pushes resources in the order of the path, transforming scattered wrong questions into a structured knowledge chain, achieving efficient and targeted delivery of targeted learning resources.

[0026] In the present embodiment, a learning effect feedback module is further included, which is configured to perform the following operations after updating the learning path each time: (a) based on the examination result data, calculate the learning effect score of the student following the previous stage learning path, which is calculated as follows: take the score rate of the target knowledge point of the current stage minus the score rate of the target knowledge point of the previous stage, and then divide by the full score value of the knowledge point and multiply by 100; wherein the score rate is defined as the ratio of the actual score of the knowledge point to the full score value of the knowledge point; (b) if the learning effect score is greater than or equal to 60 points, the learning path of the previous stage and the corresponding all learning resources are packaged as a successful case and stored to the cloud platform; (c) if the learning effect score is less than 60 points, the following remedial process is performed: extract the weight values of all knowledge points in the target student's current learning path to generate a feature sequence composed of weight values; retrieve the 5 case paths with the highest similarity to the current feature sequence from the historical cases of the cloud platform; extract the learning effect scores recorded in the 5 case paths respectively; filter out the learning resource set associated with the case with the highest learning effect score from the above cases; push the learning resource set to the terminal of the target student for reference. The setting introduces a learning effect feedback mechanism: by calculating the stage score (based on the change of knowledge point score rate), archiving and reusing successful cases, retrieving similar historical paths for low-score cases and pushing optimal resources, forming an "evaluation-archiving-optimization" closed loop, and continuously iterating to improve the effectiveness of the learning path.

[0027] In the present embodiment, the historical case retrieval process is performed according to the following steps: (i) extract all knowledge points contained in the target student's current learning path to form a current knowledge point set; (ii) retrieve the learning paths of all historical cases from the cloud platform, and extract the knowledge point set of each case path; (iii) calculate the coincidence degree of the current knowledge point set and each case knowledge point set: coincidence degree=(number of common knowledge points of the two sets) / (total amount of current knowledge point set) x 100%; (iv) select the top five case paths with the highest coincidence degree. The setting uses the coincidence degree algorithm (common knowledge point proportion) to filter similar historical cases, ensuring that the retrieval result is highly relevant to the current learning path, and providing accurate data support for remedial resource pushing.

[0028] In this embodiment, the risk warning module includes: a behavior analysis unit that identifies the following abnormal states through campus monitoring video streams: (1) abnormal mentality: represented by a continuous 3-day alone time ratio exceeding 80% and a 50% decrease in conversation frequency; (2) campus bullying: represented by physical conflicts or clothing damage in the monitoring picture, or disheveled hair and dirty clothes after entering and exiting the unmonitored area; (3) economic abnormality: represented by a continuous 5-day canteen consumption below the poverty line and a meal protein content of less than 30%; a warning response unit: sends differentiated warning information to the class teacher, security department and parents according to the type of abnormal state. This setting identifies three types of abnormal behavior (mentality, bullying, and economy) through video streams and triggers differentiated warnings (such as class teacher / security department / parents), realizes the automatic mapping from behavior data to risk level, and improves the proactive prevention and control capability of campus safety incidents.

[0029] In this embodiment, the behavior analysis unit uses multi-modal fusion recognition, specifically: a motion capture sub-unit that extracts skeletal key points based on the OpenPose algorithm to identify huddling and pushing actions; an expression recognition sub-unit that analyzes facial features through a ResNet model to output a depression index; a voice analysis sub-unit that uses an LSTM network to detect voice tremor frequency and abnormal fluctuations in volume. This setting integrates multi-modal analysis of skeletal actions (OpenPose), expressions (ResNet), and voice (LSTM), breaking through the limitations of a single data source and improving the accuracy of identifying complex abnormal states such as depression index and physical conflict.

[0030] In this embodiment, it also includes a data security module, which includes: a monitoring data desensitization unit for Gaussian blur processing of face regions in the monitoring picture; an operation audit unit for recording key operation logs through blockchain technology; a communication encryption unit for encrypting data transmission using the AES-256 algorithm. This setting provides three layers of protection through face desensitization (Gaussian blur), operation audit (blockchain), and communication encryption (AES-256) to ensure privacy security in the entire data collection, storage, and transmission process, meeting the education data compliance requirements.

[0031] In this embodiment, the notification execution engine contains an intelligent routing mechanism, specifically: teacher interview instructions are preferentially assigned to the student's free period in the last 3 days; parent notifications preferentially use phone calls for notification, and if the phone is not connected, use SMS notifications; seat adjustment instructions are associated with a three-dimensional model of the classroom to automatically generate a seating arrangement that meets the height gradient distribution and does not block the blackboard. This setting automatically assigns teacher interview periods, optimizes parent notification channels (phone → SMS), and associates with a three-dimensional model of the classroom to generate a seating arrangement based on an intelligent routing mechanism, maximizing resource scheduling efficiency, avoiding manual coordination omissions, and ensuring the rationality of seat adjustment and teaching visibility.

[0032] The application constructs an intelligent education management closed loop through multi-dimensional data analysis, and significantly improves the accuracy and timeliness of teaching intervention. Among them, the weak subject precise diagnosis is realized based on the subject ranking dynamic threshold and the score improvement weight formula (W k ) to automatically trigger differentiated intervention instructions (teacher interview / home visit). Through the four-dimensional support matching mechanism (historical relationship / subject complement / gender height / safety screening), the effectiveness of support is ensured and the seat intelligent adjustment is linked. In addition, relying on the wrong question knowledge point weight to build a dynamic learning path, combined with the historical case similarity retrieval (coincidence algorithm) to realize the adaptive optimization of learning resources. The three-level risk early warning is established by fusing multi-modal behavior recognition (skeletal action / expression / voice analysis) to prevent psychological crisis and campus bullying in real time. Through the intelligent routing engine (schedule analysis / communication priority / three-dimensional ranking), the intervention measures are automatically executed. In summary, the application can solve the problems of low efficiency of artificial analysis, lagging intervention, and inaccurate resource matching, etc. It can improve academic performance while strengthening student safety protection, and ensure data compliance through monitoring desensitization and blockchain audit.

[0033] Obviously, those skilled in the art can make various modifications and variations to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application also intends to include these modifications and variations.

[0034] The above is only the preferred specific embodiment of the present application, but the protection scope of the present application is not limited to this. Any person skilled in the art can make equivalent substitutions or changes to the technical solutions and inventive concepts of the present application within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A data analysis-based intelligent management system, characterized by, The weak subject analysis module is used for extracting subject scores of a target student from a historical examination score database, identifying current weak subjects and strong subjects of the target student, and judging whether to take remedial measures based on rank changes of the subjects in previous examinations. The helping matching module is in communication connection with the weak subject analysis module, and is used for establishing a one-to-one helping relationship of voluntary application for a student with weak subjects. The wrong question path generation module is used for analyzing wrong questions of the target student in the last three months, counting knowledge point weights, and constructing a dynamic learning path. The risk early warning module is used for detecting students with a cliff-like decline in scores, and starting a three-level risk early warning through behavior and state analysis of campus monitoring. The notification execution engine is used for triggering teacher interviews, parent communication, seat adjustment, and learning resource pushing operations according to analysis results. The weak subject analysis module comprises:

2. The intelligent management system based on data analysis according to claim 1, characterized in that, The remedial judgment unit: if the rising range of the weak subject rank in the last two examinations does not reach a threshold value, it is determined that remedial measures are taken but the effect is insufficient; if the rank continues to decline or remains unchanged, it is determined that remedial measures are not taken. The notification strategy unit: generates a teacher interview instruction when the effect is insufficient, and generates a class teacher home visit instruction when remedial measures are not taken. The helping matching module performs the following operations: Weight calculation unit: calculate the performance promotion weight W according to the formula k : ； Wherein, With is a preset adjustment coefficient, and , is the current ranking, is the previous ranking, N is the total number of students in the class, is the current score, is the previous score, and M is the full score of the subject. When W k <0.1 triggers teacher interview instruction.

3. The intelligent management system based on data analysis according to claim 2, characterized in that, (a) receiving a student voluntary helping application and archiving it to a class database; (b) when there is a target student candidate object in the database, matching is performed according to four-dimensional rules: Rule 101: Query the campus event record library. If there are records of fighting between the two parties, they are excluded. Rule 102: Analyze the subject strength-weakness comparison of the two parties. The student A's strong subject must contain the student B's weak subject, and the student B's strong subject must contain the student A's weak subject. Rule 103: Verify the gender through the student status information library. Same-sex matching is preferred. Rule 104: Compare the height values in the medical examination data. If the difference exceeds 20 cm, they are excluded. (c) After successful matching, a seat adjustment instruction is sent to the class teacher. The wrong question path generation module comprises:

4. The intelligent management system based on data analysis according to claim 3, characterized in that, The knowledge point weight calculation unit is used for counting the deduction weight of the knowledge point involved in the wrong question. The weight value calculation formula is: knowledge point deduction weight = (total deduction points of the knowledge point / total deduction points of all knowledge points) * 100%. The path construction unit takes the knowledge point with the maximum deduction weight as the starting point, links other knowledge points in order of decreasing correlation degree to form a learning path. The correlation degree is derived from the textbook knowledge graph database. The resource pushing unit pushes course videos and exercise resources to the student terminal along the learning path in order. The learning effect feedback module is used for performing the following operations after updating the learning path in each examination:

5. The intelligent management system based on data analysis according to claim 4, characterized in that, (a) Based on the examination score data, the learning effect score of the student following the previous stage learning path is calculated. The calculation method is: taking the score rate of the target knowledge point in the current stage minus the score rate of the target knowledge point in the previous stage, and then dividing by the full score value of the knowledge point and multiplying by 100; wherein the score rate is defined as the ratio of the actual score of the knowledge point to the full score value of the knowledge point. ​ (b) If the learning effect score is greater than or equal to 60 points, the learning path and all learning resources of the previous stage are packaged as a successful case and stored in the cloud platform; (c) If the learning effect score is less than 60 points, the following remedial procedures are performed: Extract the weight values of all knowledge points in the current learning path of the target student, and generate a feature sequence composed of weight values; Retrieve the top 5 case paths with the highest similarity to the current feature sequence from the historical cases in the cloud platform; Extract the learning effect scores recorded in the top 5 case paths; Select the learning resource set associated with the case with the highest learning effect score from the above cases; Push the learning resource set to the target student terminal for reference.

6. The intelligent management system based on data analysis according to claim 5, characterized in that, The historical case retrieval process is performed according to the following steps: (i) Extract all knowledge points contained in the current learning path of the target student to form a current knowledge point set; (ii) Retrieve the learning paths of all historical cases from the cloud platform, and extract the knowledge point set of each case path; (iii) Calculate the overlap between the current knowledge point set and each case knowledge point set: Overlap = (number of common knowledge points between the two sets) / (total number of current knowledge points) * 100%; (iv) Select the top five case paths with the highest overlap.

7. The intelligent management system based on data analysis according to claim 6, characterized in that, The risk warning module includes: Behavior analysis unit, which identifies the following abnormal states through campus monitoring video stream: (1) Abnormal mentality: showing a continuous 3-day alone time ratio of more than 80% and a 50% drop in conversation frequency; (2) School bullying: showing physical conflict or clothing damage in the monitoring picture, or appearing with disheveled hair and dirty clothes after entering and leaving the unmonitored area; (3) Economic anomaly: showing a continuous 5-day dining hall consumption below the poverty line and a meal protein content of less than 30%; Warning response unit: sends differentiated warning information to the class teacher, security department and parents according to the type of abnormal state.

8. The intelligent management system based on data analysis according to claim 7, characterized in that, The behavior analysis unit uses multi-modal fusion recognition, specifically: Motion capture sub-unit: extracts skeletal key points based on the OpenPose algorithm to identify crouching and pushing actions; Expression recognition sub-unit: analyzes facial features through the ResNet model to output a depression index; Voice analysis sub-unit: uses LSTM network to detect voice tremor frequency and abnormal volume fluctuations.

9. The intelligent management system based on data analysis according to claim 8, characterized in that, It also includes a data security module, which includes: Monitoring data desensitization unit for Gaussian blur processing of face regions in monitoring pictures; Operation audit unit for recording key operation logs through blockchain technology; Communication encryption unit for encrypting data transmission using AES-256 algorithm.

10. The intelligent management system based on data analysis according to claim 9, characterized in that, The notification execution engine contains an intelligent routing mechanism, specifically: Teacher interview instructions are preferentially assigned to the student's free time period in the last 3 days; Parent notification is preferentially notified by phone, and if the phone is not connected, a short message is sent; Seat adjustment instructions are associated with a classroom 3D model to automatically generate a seating arrangement that meets the height gradient distribution and does not block the blackboard.

Images