Student growth archive management method

Abstract

The invention provides a student growth archive management method, which comprises the following steps of: performing structured processing on multi-source heterogeneous data of paper archives, electronic documents and audio and video data by adopting a data acquisition and conversion technology according to heterogeneous data characteristics of student growth archive management, performing data cleaning, conversion and integration operation to obtain unified structured digital information, and storing the structured digital information into a database; a foundation is laid for subsequent big data analysis; based on a result of machine learning analysis, a student portrait and a knowledge graph are constructed, a comprehensive and dynamic student personal file is formed through multi-dimensional description of learning tracks, ability characteristics and psychological characteristics of students, entities such as students, courses, teachers and the like and relationships thereof are visualized by utilizing a knowledge graph technology, and a knowledge graph is created. The association relationship and influence factors among the students are mined, and a basis is provided for personalized education and accurate teaching.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a student growth archive management method. Background Art

[0002] Background:

[0003] When introducing big data analysis technology into student growth archive management, we face technical challenges in data collection, storage and processing. First, the sources of student archive data are diverse and scattered, including paper archives, electronic documents, audio and video materials, etc. It is necessary to establish a unified data collection and conversion mechanism to convert heterogeneous data into structured digital information. Secondly, the amount of student archive data is large and growing, which puts higher requirements on data storage and management. It is necessary to build a scalable distributed storage architecture to support efficient access and backup of massive data. Furthermore, student archive data has the characteristics of multi-dimensionality and time series. Traditional relational databases are difficult to meet complex data analysis needs. It is necessary to introduce non-relational databases and big data processing frameworks to support flexible data modeling and real-time computing. Finally, student archives involve privacy protection issues. Strict security protection measures need to be taken in various links such as data collection, transmission, storage and access to prevent data leakage and illegal access. At the same time, it is necessary to balance the contradiction between data sharing and privacy protection and maximize the value of data under the premise of legality and compliance. Summary of the invention

[0004] The present invention provides a student growth file management method, which mainly includes:

[0005] According to the heterogeneous data characteristics of student growth archive management, data collection and conversion technology is used to structure the multi-source heterogeneous data of paper archives, electronic documents, audio and video materials. Through data cleaning, conversion, and integration operations, unified structured digital information is obtained, laying the foundation for subsequent big data analysis.

[0006] In view of the large and growing volume of structured student archive data, a data storage system based on distributed storage architecture is built. The Hadoop distributed file system HDFS and NoSQL database are combined to achieve efficient storage and management of massive data. Large data sets are divided into small blocks through data sharding and distributed and stored on multiple nodes to improve parallel processing capabilities. The replica mechanism creates data copies on different nodes to ensure data reliability and availability.

[0007] In view of the multi-dimensional and time-series characteristics of the stored student file data, the non-relational databases HBase and MongoDB are introduced to support flexible data modeling and fast query. By establishing multi-dimensional indexes, such as the combined index of student ID, grade, subject and other fields, multi-angle fast retrieval is achieved. The time series index organizes data by timestamp to support efficient analysis and tracking of students' growth trajectories. At the same time, the in-memory computing engine Spark is used to achieve real-time data processing and analysis.

[0008] In the data storage, access and transmission links, strict security protection measures are taken, data encryption algorithms are used to protect the security of data in storage and transmission, role-based access control is implemented, user access rights to sensitive data are restricted, an audit log mechanism is established to record all data operation behaviors, and a data desensitization mechanism is built to achieve data sharing and utilization under the premise of protecting privacy. By formulating data management systems and processes, a balance is achieved between data sharing and privacy protection;

[0009] In view of the diversity and complexity of the student file data after security processing, we use machine learning algorithms to conduct in-depth analysis, use K-means clustering algorithm to automatically divide student groups, and divide students into different categories based on academic performance, interests and hobbies. We apply Apriori association rule mining algorithm to discover the correlation between learning behavior, extracurricular activities, interpersonal communication and other factors. Through these analyses, we can find the growth patterns and potential problems of students, and provide data support for educational decision-making;

[0010] Based on the results of machine learning analysis, we build student portraits and knowledge graphs. By characterizing students' learning trajectories, ability characteristics, and psychological characteristics in multiple dimensions, we form a comprehensive and dynamic student profile. We use knowledge graph technology to visualize entities such as students, courses, and teachers and their relationships, and explore the correlations and influencing factors between students, providing a basis for personalized education and precision teaching.

[0011] Integrate student file data with educational and teaching data, establish an educational big data analysis platform, use the student portraits and knowledge graphs constructed above, and through data visualization technology, realize the intuitive presentation and multi-dimensional analysis of students' growth status, make personalized recommendations based on student portraits, and use knowledge graphs to assist in teaching resource organization and learning path planning. Through data mining and predictive analysis, early warning and intervention are carried out on students' academic development and mental health to promote students' all-round development.

[0012] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0013] The present invention discloses a student growth archive management and analysis system. In view of the heterogeneous, massive, multi-dimensional and time-series characteristics of student archive data, the present invention adopts data acquisition and conversion technology to realize structured processing of multi-source heterogeneous data, builds a distributed storage architecture based on Hadoop and NoSQL, and introduces HBase and MongoDB to support flexible data modeling and fast query. On the basis of ensuring data security and privacy protection, the present invention uses machine learning algorithms to conduct in-depth analysis, construct student portraits and knowledge graphs, and integrate educational and teaching data to establish a big data analysis platform. Through functions such as multi-dimensional analysis, personalized recommendation and early warning intervention, the present invention realizes a comprehensive grasp of students' growth status and precise teaching, and provides strong data support and decision-making basis for promoting the all-round development of students. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 The present invention is a flow chart of a student growth file management method.

[0015] Figure 2 A schematic diagram of a student growth file management method of the present invention.

[0016] Figure 3 It is another schematic diagram of a student growth file management method of the present invention. DETAILED DESCRIPTION

[0017] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0018] like Figure 1-3 In this embodiment, a student growth file management method may specifically include:

[0019] S101. Based on the heterogeneous data characteristics of student growth archive management, data collection and conversion technology is used to structure the multi-source heterogeneous data of paper archives, electronic documents, audio and video materials. Through data cleaning, conversion, and integration operations, unified structured digital information is obtained, laying the foundation for subsequent big data analysis.

[0020] According to the heterogeneous data characteristics of student growth archive management, data collection technology is used to obtain multi-source heterogeneous data such as paper archives, electronic documents, and audio and video materials. For the collected multi-source heterogeneous data, data conversion technology is used to convert unstructured data into structured data. Through data cleaning technology, noise, duplication and invalid information in structured data are removed to improve data quality. Using data integration technology, the cleaned structured data is integrated to obtain digital information in a unified format. Feature engineering technology is used to extract key features from the integrated structured digital information to prepare for big data analysis. According to the extracted key features, appropriate machine learning algorithms, such as decision trees, support vector machines or neural networks, are selected to build prediction models. Using the constructed prediction model, the student growth archive data is analyzed and mined to discover the laws and characteristics of student growth and provide data support for educational decision-making.

[0021] Specifically, in the management of student growth archives, OCR technology can be used to scan and identify paper archives, convert images into text format, and the recognition accuracy can reach more than 95%. For electronic documents, tools such as Apache Tika can be used to extract text content. Audio and video materials can use speech recognition and video analysis technology to extract text and key information. For the collected unstructured data, natural language processing technology can be used for word segmentation, part-of-speech tagging, named entity recognition and other processing to convert it into structured digital information. In the data cleaning process, regular expressions, string matching and other methods can be used to remove duplicate records, special characters, illegal values ​​and other noise in the data to improve data quality. In the data integration stage, ETL tools such as Kettle and DataStage can be used to extract, convert and load data from different sources and formats to obtain structured data in a unified format. In the feature engineering stage, algorithms such as TF-IDF and Word2Vec can be used to extract keywords, word frequency and other features from text data, and normalize and standardize numerical data. Based on the extracted features, decision tree algorithms such as C5 and CART can be selected to build a student growth prediction model. The optimal splitting attribute can be selected through indicators such as information gain and Gini index to generate an easy-to-understand tree structure. By using the trained prediction model to analyze the student growth file data, we can find out the influence of factors such as grade, gender, and family background on student growth, and provide a quantitative basis for educational decision-making. For example, through analysis, it was found that the correlation coefficient between the academic performance in elementary school and the academic performance in high school is 78, which can intervene and help students with learning difficulties at an early stage.

[0022] S102. In view of the large and growing volume of structured student archive data, a data storage system based on a distributed storage architecture is constructed. The Hadoop distributed file system HDFS and NoSQL database are combined to achieve efficient storage and management of massive data. Large data sets are divided into small blocks through data sharding and distributed and stored on multiple nodes to improve parallel processing capabilities. The replica mechanism creates data copies on different nodes to ensure data reliability and availability.

[0023] According to the characteristics of student archive data, a distributed storage architecture is adopted to combine HDFS and NoSQL database to achieve efficient storage and management of massive data. Through data sharding technology, large student archive data sets are divided into small blocks and distributed and stored on multiple data nodes of HDFS to improve the parallel ability of data processing. In HDFS, a copy mechanism is adopted to copy each data block to different data nodes to ensure data reliability and availability. In view of the structured characteristics of student archive data, NoSQL database is used for data storage and query to improve the flexibility and performance of data access. Through the MapReduce parallel computing framework, batch processing and analysis of student archive data distributed on HDFS are carried out to achieve efficient use of data. Data compression technology is used to compress student archive data stored in HDFS and NoSQL databases to reduce storage space and improve data transmission efficiency. Through data backup and fault tolerance mechanisms, student archive data in HDFS and NoSQL databases are backed up regularly to ensure data security and recoverability and improve the fault tolerance of the system.

[0024] Specifically, for the storage and management of massive student archive data, a distributed storage architecture is adopted to combine HDFS and NoSQL database. Through data sharding technology, the student archive data set is divided into 128MB data blocks according to the hash value of the student ID, and distributed and stored on an HDFS cluster consisting of 10 data nodes. Each data block is replicated 3 times and stored on different data nodes to ensure data reliability and availability. In view of the structured characteristics of student archive data, MongoDB is used as a NoSQL database to store student basic information, student status information, and grade information in different collections, and associate them through student IDs to improve the flexibility of data access and query performance. The MapReduce parallel computing framework is used to batch process and analyze student archive data distributed on HDFS, such as calculating the average score of each grade and counting the number of students in each major. By assigning computing tasks to multiple nodes for parallel execution, data processing efficiency is achieved. The Snappy compression algorithm is used to compress student archive data stored in HDFS and MongoDB, and the compression ratio can reach 3:1, effectively reducing storage space usage. By setting up data backup strategies, the student file data in HDFS and MongoDB is fully backed up at 2 a.m. every day. The backup data is stored in an independent backup cluster. At the same time, the incremental backup mechanism is enabled to perform incremental backups of new and modified data every 6 hours to ensure data security and recoverability. Through the comprehensive application of these technical means, the efficient storage, management and utilization of massive student file data is achieved, providing strong data support for the school's teaching management and decision-making analysis.

[0025] S103. In view of the multi-dimensional and time-series characteristics of the stored student file data, the non-relational databases HBase and MongoDB are introduced to support flexible data modeling and fast query. By establishing multi-dimensional indexes, such as the combined index of fields such as student ID, grade, and subject, multi-angle fast retrieval is achieved. The time series index organizes data by timestamp to support efficient analysis and tracking of students' growth trajectories. At the same time, the in-memory computing engine Spark is used to achieve real-time data processing and analysis.

[0026] According to the multi-dimensional attributes of student archive data, HBase and MongoDB non-relational databases are used for flexible data modeling and storage, supporting dynamic dimension addition. For key fields such as student ID, grade, and subject, multi-dimensional combined indexes are established to quickly filter and retrieve target data through indexes to improve query performance. The timestamp information of student archive data is obtained, the data is organized and stored according to time series, and a time-based index is constructed to achieve efficient analysis and tracking of student growth trajectories. If massive student archive data needs to be processed and analyzed in real time, the Spark memory computing engine is used to load data into memory for iterative calculations to improve data processing speed. Structured query analysis of student archive data is performed through Spark SQL, which supports flexible ad hoc queries and meets the needs of multi-angle data analysis. Machine learning algorithms such as clustering and association rule mining are used to perform cluster analysis on student archive data to discover student group characteristics and laws, providing decision support for personalized teaching. The multi-dimensional attribute data of student archives is vectorized, similar students are found through similarity calculation, and student portraits and accurate recommendations are achieved based on collaborative filtering algorithms.

[0027] S104. Take strict security measures in the data storage, access and transmission links. Use data encryption algorithms to protect the security of data in storage and transmission. Implement role-based access control to limit user access to sensitive data. Establish an audit log mechanism to record all data operations. At the same time, build a data desensitization mechanism to achieve data sharing and utilization under the premise of protecting privacy. Balance data sharing and privacy protection by formulating data management systems and processes.

[0028] According to the sensitivity of the data, encryption algorithms of different strengths are used to encrypt the data to obtain the encrypted data. Through the access control policy, the user role and permissions are judged. If the user has the corresponding permissions, the permission to access the encrypted data is granted. When the data is accessed or operated, the audit log function is triggered to record the user, time, type and other information of the data operation to form a data operation audit log. The original data is obtained, and the sensitive information is removed through the data desensitization algorithm to obtain the desensitized data for sharing and analysis. A data classification management system is formulated to classify the data according to the sensitivity level of the data, and the corresponding management measures and processes are determined. Privacy-preserving machine learning algorithms such as differential privacy are used to achieve data aggregation analysis and mining without leaking individual privacy. A data security management process is established to incorporate data encryption, access control, auditing, desensitization and other links into the management process to form a full-link data security guarantee mechanism.

[0029] Specifically, for highly sensitive data, high-strength encryption algorithms such as AES-256 can be used, while for general sensitive data, relatively low-strength algorithms such as AES-128 can be used. Through the role-based access control (RBAC) strategy, users are divided into different roles such as administrators, operators, auditors, and corresponding data access permissions are set for each role. When a user requests to access encrypted data, the system automatically determines the role and permission to which it belongs, and only authorized users are allowed to access. When accessing or operating data, the system automatically records information such as the operating user's IP, timestamp, and operation type, and stores it in the form of a log with a retention period of 6 months. For sensitive information such as ID card number and mobile phone number in the original data, data desensitization algorithms such as masking and encryption are used for processing. For example, the ID card number "110101198001010019" is desensitized to "1101**********0019" before data sharing and analysis. According to the sensitivity of the data, the data is divided into three levels: core business secrets, internal sensitive information, and externally shareable data. Different management measures are taken for data of different levels. For example, core business secrets must be stored in an offline environment and managed by a dedicated person. Using the differential privacy machine learning algorithm, random noise is introduced during data aggregation analysis, so that the privacy of individual individuals is protected, but the overall data distribution characteristics are not affected. Incorporate data security management into the enterprise information security management system (ISMS), formulate data security management systems and emergency plans, and realize data full life cycle management through process-based IT systems to form a traceable and auditable data security closed loop.

[0030] S105. In view of the diversity and complexity of the student file data after security processing, a machine learning algorithm is used for in-depth analysis. The K-means clustering algorithm is used to automatically divide the student groups and classify students into different categories based on their academic performance, interests and hobbies. The Apriori association rule mining algorithm is used to discover the correlation between learning behavior, extracurricular activities, interpersonal communication and other factors. Through these analyses, the growth patterns and potential problems of students are discovered, providing data support for educational decision-making.

[0031] According to the student file data, data preprocessing technology is used to process missing values, outliers, etc. to ensure data quality. For the preprocessed student file data, feature engineering methods are used to extract key features such as academic performance and hobbies to construct student feature vectors. According to the student feature vectors, the K-means clustering algorithm is used to divide students into different groups by calculating the distance between samples to obtain student classification results. For the clustered student groups, the Apriori association rule mining algorithm is used to mine student behavior patterns by analyzing the correlation between factors such as learning behavior, extracurricular activities, and interpersonal communication. According to the association rule mining results, the potential problems of students in learning, life, etc. are judged, and the student groups that need attention and help are identified. For the identified potential problems of students, the decision tree algorithm is used to generate educational decision rules by analyzing the relationship between student characteristics and problems. According to the educational decision rules, personalized education plans are formulated for different student groups to promote the healthy growth of students and improve the quality of teaching.

[0032] Specifically, the student file data is first preprocessed, the maximum likelihood estimation method is used to fill the missing values, and the box plot and Z-score method are used to identify and process the outliers to ensure the integrity and accuracy of the data. Then, the feature engineering method is used to select 10 key features such as academic performance and hobbies through Pearson correlation coefficient calculation, and the Min-Max normalization method is used to scale the features to construct the student feature vector. Then, the K-means clustering algorithm is used to calculate the similarity between samples through Euclidean distance, set the number of clusters K=4, and iterate 20 times to obtain the student classification results, forming four groups of strong learning ability, medium learning ability, weak learning ability and poor learning ability. For the clustered student groups, the Apriori association rule mining algorithm is used, with the minimum support set to 05 and the minimum confidence set to 8. The behavior patterns of students with strong learning ability generally participate in science and technology clubs and read extracurricular books, while students with weak learning ability participate more in physical exercise and entertainment activities. According to the association rule mining results, it is judged that students with poor learning ability have potential problems in learning methods and time management, and need to be focused on and helped. For students with weak learning ability and poor learning ability, the C5 decision tree algorithm is used to select split attributes through information gain ratio to generate educational decision rules. For example, for students with academic performance below 60 points and extracurricular reading time less than 1 hour / day, a daily study plan needs to be formulated and supervised and guided by the class teacher. Finally, according to the educational decision rules, personalized education plans are formulated for different student groups. For example, students with strong learning ability are trained to be outstanding, encouraged to participate in subject competitions and publish papers, and students with weak learning ability are guided in learning methods and psychological counseling to promote their all-round development and healthy growth. Through the application of data mining technology, student portraits and precise teaching are realized, which effectively improves the teaching quality and management level.

[0033] S106. Construct student portraits and knowledge graphs based on the results of machine learning analysis. Create a comprehensive and dynamic student profile by characterizing students’ learning trajectories, ability characteristics, and psychological characteristics in multiple dimensions. Use knowledge graph technology to visualize entities such as students, courses, and teachers and their relationships, and explore the associations and influencing factors between students, providing a basis for personalized education and precision teaching.

[0034] According to the above business content and the extracted related attributes, the interactive data of the student learning platform is collected to obtain the learning trajectory information such as the student's online learning time, homework submission frequency, and test scores. The clustering algorithm is used to segment and classify the students according to the learning trajectory information to determine the learning mode characteristics of different student types. The text information submitted by students on the platform is analyzed by natural language processing technology. The key words reflecting the ability characteristics and psychological characteristics are extracted through word frequency statistics and sentiment analysis. The key words are integrated and analyzed using the topic model to construct a structured label system of ability characteristics and psychological characteristics. According to the subdivision type of student learning trajectory and the corresponding learning mode characteristics, the corresponding labels in the student ability characteristics and psychological characteristics label system are matched. The association rule mining algorithm is used to analyze the implicit connection between the learning mode characteristics and the ability characteristics and psychological characteristics labels, and the preliminary framework of the student portrait is obtained. The syllabus and assessment standards of all courses are summarized, the knowledge points related to the courses and the dependencies between the knowledge points are obtained, the nodes and edges of the course knowledge graph are constructed, the attribute information of each knowledge point is marked, and the construction and storage of the course knowledge graph are completed. Obtain the teacher's personal basic information and research direction, analyze the course knowledge graph of the course taught by the teacher, use the PageRank algorithm to quantitatively evaluate the teacher's influence, determine the teacher's core node in the knowledge graph and its associated knowledge field based on the impact evaluation results and the teacher's research direction, and integrate the teacher entity into the course knowledge graph. According to the student portrait framework, integrate the student's personal basic information and learning trajectory segmentation types, map the ability characteristics and psychological characteristic labels and corresponding knowledge point annotations to the student entity in the knowledge graph, use the graph embedding algorithm to convert students, courses, teachers and their mutual relationships into vector representations, calculate the similarity of student entities in the graph, and judge the potential correlation and influencing factors between students. According to the students' correlation and influencing factors in the knowledge graph, combined with the students' ability characteristics and psychological characteristic labels, apply the recommendation algorithm to match personalized learning resources and adaptive teaching strategies, generate personalized teaching plans including learning path planning and tutoring plans, and apply them to actual teaching links.

[0035] Specifically, students' interactive data are collected through the learning platform, including learning trajectory information such as learning time, homework submission frequency, and test scores. The K-means clustering algorithm is used, and the number of clusters K is set to 4, and the number of iterations is 10. Students are clustered according to the dimensions of student learning time, homework submission frequency, and test scores. Four types of students are obtained, including those with positive learning attitude, strong learning ability, general learning attitude, and weak learning ability. The learning pattern characteristics of each type of students are analyzed. Jieba word segmentation and Word2Vec word vector are used to segment and vectorize the text information submitted by students, and high-frequency words are counted. In combination with the SnowNLP sentiment analysis tool, the sentiment tendency score of the text is calculated, and keywords reflecting students' ability characteristics and psychological characteristics are extracted. The LDA topic model is used, and the number of topics is set to 5 and the number of iterations is 500. Keywords are clustered by topic, and a student ability and psychological characteristic label system with five dimensions of learning ability, thinking ability, emotional management, interpersonal communication, and self-cognition is constructed. Based on the Apriori association rule mining algorithm, the minimum support is set to 2 and the minimum confidence is set to 5. The association rules between the characteristics of students' learning patterns and their psychological labels of abilities are analyzed to obtain students' abilities and psychological portraits under different learning patterns. The course syllabus and assessment standards are sorted out, the knowledge points of each course are extracted, and the course knowledge graph is constructed. The knowledge points are taken as nodes, the dependencies between the knowledge points are directed edges, and the names, courses, importance, and difficulty of mastering of the knowledge points are marked. The Neo4j graph database is used to construct and store the knowledge graph. The personal information, teaching records, and research directions of teachers are obtained, and teachers are used as entity nodes in the knowledge graph. The PageRank algorithm is used to calculate the centrality of the teacher nodes. The number of iterations is set to 100 times. The teachers with the top 20% of the centrality are defined as core nodes. The research fields of teachers are associated with the teaching knowledge points, and the knowledge graph is optimized by integrating teacher information. The student portrait information is mapped to the knowledge graph, and the student abilities and psychological labels correspond to the relevant knowledge points and form entity associations. The DeepWalk graph embedding algorithm is used, and the number of random walk steps is set to 10, the window size is set to 5, the number of iterations is set to 50, and the learning rate is set to 0.25. The entity relationship between students, courses, and teachers is embedded into a 100-dimensional vector space. The similarity of student nodes is calculated by cosine similarity. Students exceeding the threshold of 8 are determined as potential related groups, and the influencing factors between students are inferred. Finally, the student portrait features and knowledge graph information are combined, and the content-based collaborative filtering recommendation algorithm is used to match personalized learning resources for students. According to the students' current knowledge mastery and learning progress, the A* search algorithm is used to plan the optimal learning path in the knowledge graph, and based on the students' ability characteristics and psychological characteristics, a targeted learning tutoring plan is given to form a complete personalized teaching plan for teaching practice.

[0036] S107. Integrate student archive data with educational and teaching data to establish an educational big data analysis platform. Using the student portraits and knowledge graphs constructed above, through data visualization technology, intuitive presentation and multi-dimensional analysis of student growth status can be achieved. Personalized recommendations are made based on student portraits, and knowledge graphs are used to assist in the organization of teaching resources and learning path planning. Through data mining and predictive analysis, early warning and intervention are carried out on students' academic development and mental health to promote the all-round development of students.

[0037] According to student files and teaching data, data cleaning and data integration technologies are used to uniformly process heterogeneous data to obtain standardized data sets. According to standardized data sets, clustering algorithms and association rule mining algorithms are used to analyze students' learning behavior patterns and knowledge mastery to obtain student portraits and knowledge graphs. According to student portraits, collaborative filtering recommendation algorithms are used to recommend personalized teaching resources and learning paths for students in combination with students' learning interests and knowledge mastery. According to knowledge graphs, teaching resources are organized and associated through semantic analysis technology to form a structured knowledge system to assist students in systematic learning. According to students' learning process data, data visualization technology is used to generate a multi-dimensional analysis report on students' learning status, which intuitively presents students' growth trajectory and knowledge mastery. According to students' historical learning data, time series prediction algorithms are used to predict students' future academic development trends and timely discover learning difficulties and potential problems. According to students' behavioral data and emotional data, sentiment analysis technology and anomaly detection algorithms are used to evaluate and warn students' mental health status to promote students' physical and mental health development.

[0038] Specifically, in the data cleaning and integration stage, ETL tools such as Talend or Informatica can be used to extract, convert and load heterogeneous data such as student files and teaching data, unify data formats and encodings, remove duplicate and invalid data, and form a standardized data set. Then, the K-means clustering algorithm is used to divide students into 5 categories according to their learning behavior patterns, and then the Apriori association rule mining algorithm is used to find out the knowledge mastery characteristics of each type of students, with the support set to 20% and the confidence set to 80%, so as to obtain detailed student portraits and knowledge graphs. Based on this, the item-based collaborative filtering algorithm is applied to calculate the student similarity based on the student's learning interest and knowledge mastery, and recommend the Top-N personalized teaching resources and learning paths for each student. At the same time, semantic analysis technologies such as Word2Vec are used to vectorize teaching resources, and related resources are organized into a structured knowledge system through cluster analysis to facilitate students' systematic learning. In addition, data visualization tools such as Tableau or PowerBI are used to convert students' learning process data such as learning time and exercise scores into intuitive charts to present students' growth trajectories from multiple angles. For students' historical performance data, we use time series prediction models such as LSTM, with previous performance as feature variables and next performance as target variables, to train prediction models with an accuracy rate of more than 85%, providing timely warnings of academic risks. Finally, we use sentiment dictionaries and machine learning algorithms to analyze the sentiment tendency of student logs and discussion content, and then use anomaly detection algorithms such as isolation forests to identify students with abnormal emotions, forming a mental health warning model with an accuracy rate of 90%, promoting the all-round development of students.

[0039] The description of the above embodiments is only used to help understand the technical solutions and core ideas of the present application. Ordinary technical personnel in this field should understand that they can still modify the technical solutions recorded in the aforementioned embodiments, or replace some of the technical features therein with equivalents. However, these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A student growth file management method, characterized in that: The method comprises: According to the heterogeneous data characteristics of student growth archive management, data collection and conversion technology is used to structure the multi-source heterogeneous data of paper archives, electronic documents, audio and video materials. Through data cleaning, conversion, and integration operations, unified structured digital information is obtained, laying the foundation for subsequent big data analysis. In view of the large and growing volume of structured student archive data, a data storage system based on distributed storage architecture is built. The Hadoop distributed file system HDFS and NoSQL database are combined to achieve efficient storage and management of massive data. Large data sets are divided into small blocks through data sharding and distributed and stored on multiple nodes to improve parallel processing capabilities. The replica mechanism creates data copies on different nodes to ensure data reliability and availability. In view of the multi-dimensional and time-series characteristics of the stored student file data, the non-relational databases HBase and MongoDB are introduced to support flexible data modeling and fast query. By establishing multi-dimensional indexes, such as the combined index of student ID, grade, subject and other fields, multi-angle fast retrieval is achieved. The time series index organizes data by timestamp to support efficient analysis and tracking of students' growth trajectories. At the same time, the in-memory computing engine Spark is used to achieve real-time data processing and analysis. In the data storage, access and transmission links, strict security protection measures are taken, data encryption algorithms are used to protect the security of data in storage and transmission, role-based access control is implemented, user access rights to sensitive data are restricted, an audit log mechanism is established to record all data operation behaviors, and a data desensitization mechanism is built to achieve data sharing and utilization under the premise of protecting privacy. By formulating data management systems and processes, a balance is achieved between data sharing and privacy protection; In view of the diversity and complexity of the student file data after security processing, we use machine learning algorithms to conduct in-depth analysis, use K-means clustering algorithm to automatically divide student groups, and divide students into different categories based on academic performance, interests and hobbies. We apply Apriori association rule mining algorithm to discover the correlation between learning behavior, extracurricular activities, interpersonal communication and other factors. Through these analyses, we can find the growth patterns and potential problems of students, and provide data support for educational decision-making; Based on the results of machine learning analysis, we build student portraits and knowledge graphs. By characterizing students' learning trajectories, ability characteristics, and psychological characteristics in multiple dimensions, we form a comprehensive and dynamic student profile. We use knowledge graph technology to visualize entities such as students, courses, and teachers and their relationships, and explore the correlations and influencing factors between students, providing a basis for personalized education and precision teaching. Integrate student file data with educational and teaching data, establish an educational big data analysis platform, use the student portraits and knowledge graphs constructed above, and through data visualization technology, realize the intuitive presentation and multi-dimensional analysis of students' growth status, make personalized recommendations based on student portraits, and use knowledge graphs to assist in teaching resource organization and learning path planning. Through data mining and predictive analysis, early warning and intervention are carried out on students' academic development and mental health to promote students' all-round development.

2. The method according to claim 1, characterized in that According to the heterogeneous data characteristics of student growth archive management, data collection and conversion technology is used to structure the multi-source heterogeneous data of paper archives, electronic documents, audio and video materials; through data cleaning, conversion, and integration operations, unified structured digital information is obtained, laying the foundation for subsequent big data analysis, including: According to the heterogeneous data characteristics of student growth archive management, data collection technology is used to obtain multi-source heterogeneous data such as paper archives, electronic documents, and audio and video materials; For the collected multi-source heterogeneous data, data conversion technology is used to convert unstructured data into structured data; Through data cleaning technology, remove noise, duplication and invalid information in structured data to improve data quality; Use data integration technology to integrate cleaned structured data to obtain digital information in a unified format; Use feature engineering techniques to extract key features from integrated structured digital information to prepare for big data analysis; Based on the extracted key features, select appropriate machine learning algorithms, such as decision trees, support vector machines, or neural networks, to build prediction models; The constructed predictive model is used to analyze and mine student growth file data to discover the patterns and characteristics of student growth and provide data support for educational decision-making.

3. The method according to claim 1, characterized in that In view of the fact that the amount of structured student archive data is large and growing, a data storage system based on a distributed storage architecture is constructed; a combination of Hadoop distributed file system HDFS and NoSQL database is adopted to achieve efficient storage and management of massive data; large data sets are divided into small blocks through data sharding, and distributed and stored on multiple nodes to improve parallel processing capabilities; The replication mechanism creates data copies on different nodes to ensure data reliability and availability, including: According to the characteristics of student file data, a distributed storage architecture is adopted to combine HDFS and NoSQL database to achieve efficient storage and management of massive data; Through data sharding technology, large student archive data sets are divided into small blocks and distributed and stored on multiple data nodes of HDFS to improve the parallel capability of data processing; In HDFS, a replication mechanism is used to copy each data block to different data nodes to ensure data reliability and availability; In view of the structured characteristics of student file data, NoSQL database is used for data storage and query to improve the flexibility and performance of data access; Through the MapReduce parallel computing framework, batch processing and analysis of student file data distributed on HDFS are carried out to achieve efficient use of data; Use data compression technology to compress student file data stored in HDFS and NoSQL databases, reduce storage space usage, and improve data transmission efficiency; Through data backup and fault tolerance mechanisms, student file data in HDFS and NoSQL databases are backed up regularly to ensure data security and recoverability and improve the system's fault tolerance.

4. The method according to claim 1, characterized in that In view of the multi-dimensional and time-series characteristics of the stored student file data, the non-relational databases HBase and MongoDB are introduced to support flexible data modeling and fast query; multi-dimensional indexes are established, such as the combined index of fields such as student ID, grade, and subject, to achieve multi-angle fast retrieval; time series indexes organize data by timestamp to support efficient analysis and tracking of student growth trajectories; and the in-memory computing engine Spark is used to achieve real-time data processing and analysis, including: Based on the multi-dimensional attributes of student file data, HBase and MongoDB non-relational databases are used for flexible data modeling and storage, supporting dynamic addition of dimensions; For key fields such as student ID, grade, and subject, we build multi-dimensional combined indexes to quickly filter and retrieve target data through indexes, thus improving query performance. Obtain the timestamp information of student file data, organize and store data according to time series, build a time-based index, and achieve efficient analysis and tracking of student growth trajectories; If it is necessary to process and analyze massive amounts of student file data in real time, the Spark memory computing engine is used to load the data into memory for iterative calculations to increase data processing speed; Use SparkSQL to perform structured query analysis on student file data, support flexible instant query, and meet multi-angle data analysis needs; Use machine learning algorithms such as clustering and association rule mining to perform cluster analysis on student file data, discover student group characteristics and patterns, and provide decision support for personalized teaching; The multi-dimensional attribute data of student files are vectorized, similar students are found through similarity calculation, and student profiling and accurate recommendations are achieved based on collaborative filtering algorithms.

5. The method according to claim 1, characterized in that Strict security measures are taken in the data storage, access and transmission links; data encryption algorithms are used to protect the security of data in storage and transmission; role-based access control is implemented to limit user access rights to sensitive data; Establish an audit log mechanism to record all data operations; At the same time, a data desensitization mechanism is built to achieve data sharing and utilization under the premise of protecting privacy; Balance data sharing and privacy protection by developing data management systems and processes, including: According to the sensitivity of the data, encryption algorithms of different strengths are used to encrypt the data to obtain encrypted data; The user role and permissions are determined through access control policies. If the user has the corresponding permissions, he is granted access to the encrypted data. When data is accessed or operated, the audit log function is triggered to record the user, time and type of data operation information to form a data operation audit log; Obtain the original data, remove sensitive information through the data desensitization algorithm, and obtain the desensitized data for sharing and analysis; Formulate a data classification management system, classify data according to its sensitivity level, and determine corresponding management measures and processes; Adopt privacy-preserving machine learning algorithms such as differential privacy to achieve data aggregation analysis and mining without leaking individual privacy; Establish a data security management process, incorporate data encryption, access control, auditing and desensitizing into the management process, and form a full-link data security protection mechanism.

6. The method according to claim 1, characterized in that In view of the diversity and complexity of the student file data after security processing, a machine learning algorithm is used for in-depth analysis; a K-means clustering algorithm is used to automatically divide the student group and divide the students into different categories based on characteristics such as academic performance and interests; an Apriori association rule mining algorithm is applied to discover the correlation between factors such as learning behavior, extracurricular activities, and interpersonal communication; Through these analyses, we can discover students’ growth patterns and potential problems, and provide data support for educational decision-making, including: According to the student file data, data preprocessing technology is used to process missing values ​​and outliers to ensure data quality; Based on the preprocessed student file data, feature engineering methods are used to extract key features such as academic performance and interests and hobbies, and construct student feature vectors; According to the student feature vector, the K-means clustering algorithm is used to divide the students into different groups by calculating the distance between samples, and the student classification results are obtained; For the clustered student groups, the Apriori association rule mining algorithm is used to mine student behavior patterns by analyzing the correlation between factors such as learning behavior, extracurricular activities, and interpersonal communication; Based on the association rule mining results, we can judge the potential problems of students in their studies and life, and identify the groups of students who need attention and help; For the identified potential problems of students, a decision tree algorithm is used to generate educational decision rules by analyzing the relationship between student characteristics and problems; According to the educational decision-making rules, personalized education plans are formulated for different student groups to promote the healthy growth of students and improve the quality of teaching.

7. The method according to claim 1, characterized in that Based on the results of machine learning analysis, student portraits and knowledge graphs are constructed; comprehensive and dynamic student profiles are formed by multi-dimensionally characterizing students' learning trajectories, ability characteristics, and psychological characteristics; knowledge graph technology is used to visualize entities such as students, courses, and teachers and their relationships, and to explore the associations and influencing factors between students, providing a basis for personalized education and precision teaching, including: According to the above business content and the extracted related attributes, the interactive data of the student learning platform is collected to obtain the learning trajectory information such as the online learning time, homework submission frequency and test scores of the students. The clustering algorithm is used to segment and classify the students according to the learning trajectory information to determine the learning mode characteristics of different student types; Use natural language processing technology to analyze the text information submitted by students on the platform, extract key words that reflect ability characteristics and psychological characteristics through word frequency statistics and sentiment analysis, use topic models to integrate and analyze key words, and build a structured label system for ability characteristics and psychological characteristics; According to the subdivision types of students' learning trajectories and the corresponding learning mode characteristics, the students' ability characteristics are matched with the corresponding labels in the psychological characteristic label system. The association rule mining algorithm is used to analyze the implicit connection between the learning mode characteristics and the ability characteristics and psychological characteristic labels, and the preliminary framework of the student portrait is obtained. Summarize the syllabus and assessment standards of all courses, obtain the knowledge points associated with the courses and the dependencies between the knowledge points, construct the nodes and edges of the course knowledge graph, annotate the attribute information of each knowledge point, and complete the construction and storage of the course knowledge graph; Obtain the teacher's basic personal information and research direction, analyze the course knowledge graph of the course taught by the teacher, use the PageRank algorithm to quantitatively evaluate the teacher's influence, and determine the teacher's core node in the knowledge graph and its related knowledge fields based on the influence evaluation results and the teacher's research direction, and integrate the teacher entity into the course knowledge graph; Based on the student portrait framework, the basic personal information of students and the subdivision types of learning trajectories are integrated, and the ability characteristics and psychological characteristic labels and corresponding knowledge point annotations are mapped to the student entities in the knowledge graph. The graph embedding algorithm is used to convert students, courses and teachers and their relationships into vector representations, and the similarity of student entities in the graph is calculated to determine the potential correlation and influencing factors between students. Based on the associations and influencing factors of students in the knowledge graph, combined with students' ability characteristics and psychological feature labels, the recommendation algorithm is applied to match personalized learning resources and adaptive teaching strategies, and generate personalized teaching plans that include learning path planning and tutoring plans, which are applied to actual teaching links.

8. The method according to claim 1, characterized in that The above-mentioned integration of student archive data and educational teaching data to establish an educational big data analysis platform; using the student portraits and knowledge graphs constructed above, through data visualization technology, to achieve intuitive presentation and multi-dimensional analysis of student growth status; based on student portraits, personalized recommendations are made, and knowledge graphs are used to assist in the organization of teaching resources and learning path planning; Through data mining and predictive analysis, early warning and intervention are carried out on students’ academic development and mental health to promote their all-round development, including: Based on student files and teaching data, data cleaning and data integration technology are used to uniformly process heterogeneous data to obtain standardized data sets; Based on the standardized data set, clustering algorithm and association rule mining algorithm are used to analyze students' learning behavior patterns and knowledge mastery, and obtain student portraits and knowledge graphs; Based on student portraits, collaborative filtering recommendation algorithms are used to recommend personalized teaching resources and learning paths to students in combination with their learning interests and knowledge mastery. According to the knowledge graph, through semantic analysis technology, teaching resources are organized and associated to form a structured knowledge system to assist students in systematic learning; Based on the students' learning process data, data visualization technology is used to generate a multi-dimensional analysis report on the students' learning status, which intuitively presents the students' growth trajectory and knowledge mastery; Based on students' historical learning data, time series prediction algorithms are used to predict students' future academic development trends and identify learning difficulties and potential problems in a timely manner; Based on students' behavioral and emotional data, sentiment analysis technology and anomaly detection algorithms are used to evaluate and warn students' mental health status, thereby promoting their healthy physical and mental development.

Images