Teaching big data resource management system based on cloud database

Abstract

The invention relates to the technical field of big data, and discloses a teaching big data resource management system based on a cloud database, the system integrates a data servicing unit, a cloud base unit and an intelligent center unit through a layered distributed architecture, a closed-loop service system is constructed, and the system depends on the elastic storage capability of the cloud database to realize resource management. A teaching scene load is sensed in real time through an elastic scheduling module, database computing resources and storage space are dynamically adjusted by using a storage resource expansion model and a resource dynamic allocation function, automatic capacity expansion in a high-load period and accurate capacity reduction in a low-load period are realized, the problems of performance bottleneck and resource idleness of a traditional scheme are avoided, and the teaching efficiency is improved. And meanwhile, data security is guaranteed through cross-regional synchronization, encrypted storage and disaster recovery backup, intelligent analysis and recommendation of teaching big data are realized in combination with a knowledge graph and machine learning, elastic and efficient cloud database service and big data resource service management capability are provided for a teaching scene, and resource utilization efficiency and teaching service quality are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of big data, and particularly relates to a teaching big data resource management system based on a cloud database. BACKGROUND

[0002] Under the wave of education informatization and digital transformation, diversified teaching scenarios such as online teaching, smart classrooms and digital evaluation are booming, which has led to the continuous accumulation and dynamic update of massive teaching data. These data cover various types such as course videos, electronic teaching plans and learning behavior records, and include not only structured achievement data but also unstructured classroom voice and blackboard image resources, which puts forward higher requirements for data integration management, storage architecture and application services. Cloud database provides technical support for efficient governance of teaching big data with its characteristics of elastic expansion, high availability and distributed storage. Building a resource management system integrating cloud database capabilities is a key path to realize standardized preparation, intelligent analysis and large-scale application of teaching data, and can effectively promote the digital integration and value release of educational resources.

[0003] Patent No. CN108647265A discloses a multi-platform data interactive system, which extracts data through a bottom integrated platform and is applied to data resource management and big data application support platforms. The resource management platform extracts, converts and displays SOLR, HBASE, ORACLE and GP multi-source structured data in list form with the help of an extractor. The big data application support platform takes user service as the core, assigns differentiated permissions according to roles, realizes different roles to view exclusive data services, and provides a subscription function to meet the personalized data management and use needs of users.

[0004] For the above and existing related technologies, the inventors believe that the following defects often exist: the existing big data resource management system cannot dynamically adjust cloud database storage and computing resources according to teaching scenario load, leading to performance bottlenecks during peak periods, slow response, service interruption and other problems, which seriously affect the teaching experience; resources are idle during low load, causing resource waste and increasing operating costs, and it is difficult to meet the needs of cross-regional collaboration and secure storage of teaching resources. SUMMARY

[0005] The technical problem to be solved by the present application is that the existing big data resource management system cannot dynamically adjust cloud database storage and computing resources according to teaching scenario load, leading to performance bottlenecks during peak periods and idle resources during low load. Therefore, we propose a teaching big data resource management system based on a cloud database.

[0006] In order to achieve the above objectives, this application adopts the following technical solutions: a teaching big data resource management system based on cloud database, comprising:

[0007] Data preparation unit: responsible for unified access, standardized description, and decomposition of raw data from different teaching scenarios into manageable knowledge units, laying the foundation for subsequent processing;

[0008] Cloud base unit: used to provide elastically scalable cloud storage architecture and infrastructure support, specifically:

[0009] Resource pool module: used to provide centralized and standardized cloud storage services, centrally manage all prepared teaching resource data, and form an efficient access data pool;

[0010] Elastic Scheduling Module: This module dynamically allocates and adjusts the computing power, storage space, and network bandwidth resources of the cloud database based on real-time access load to ensure stable system performance.

[0011] Cross-region synchronization module: used to achieve real-time or near-real-time data synchronization of teaching resources between nodes in different regions of the cloud, and implement cross-campus or cross-institutional educational resource sharing and collaboration;

[0012] Encrypted storage module: used to apply encryption technology on data transmission links and static storage media to protect the confidentiality of sensitive data in teaching resources in the cloud; and

[0013] Disaster recovery module: This module implements redundant storage of multiple copies of data through a distributed storage architecture and executes regular backup strategies to ensure data persistence and disaster recovery capabilities.

[0014] Intelligent central unit: used to conduct in-depth analysis and mining of teaching data through the use of knowledge graphs, user portraits and machine learning algorithms to drive intelligent services.

[0015] Preferably, it also includes a teaching diagnosis module and a system operation and maintenance unit:

[0016] Teaching Diagnosis Module: This module is used to comprehensively analyze multi-dimensional information including classroom interaction data, homework completion, and test scores to evaluate the overall teaching effect and identify key teaching difficulties.

[0017] System Operation and Maintenance Unit: This unit is used to ensure the safe and compliant operation of the entire system, authority control, and continuous performance monitoring. It is divided into:

[0018] Permission control module: used to implement role-based access control strategies, define the operation permissions of different user roles for various teaching resources, and ensure data security and privacy protection;

[0019] System monitoring module: used to collect, analyze and visualize the system's key performance indicators, resource usage status and operation logs in real time, providing a decision-making basis for system operation and maintenance, optimization and troubleshooting.

[0020] Preferably, the data preparation unit includes:

[0021] Multi-source access module: This module is responsible for connecting and aggregating structured and unstructured data sources generated by teaching scenarios, including online courses, classroom recordings, homework systems, and examination platforms, to achieve unified data entry and preliminary cleaning.

[0022] Metadata management module: responsible for formulating, maintaining and implementing unified resource description metadata standards to ensure that all teaching resources have complete, consistent and interoperable metadata information;

[0023] Resource fragmentation module: used to use natural language processing and machine learning technologies to intelligently break down complex teaching content into fine-grained knowledge slices or units, and establish their association with knowledge points and user behavior.

[0024] Preferably, the flexible scheduling module specifically includes:

[0025] Load sensing module: used to collect CPU, memory, network bandwidth, and storage IOPS indicators of each node in the teaching system in real time, predict resource demand trends based on business scenarios, and provide data support for dynamic scheduling;

[0026] Policy decision module: Used for the built-in rule engine and machine learning model to formulate resource allocation strategies based on load data;

[0027] Compute scaling module: used to connect to the Serverless computing service, dynamically create or release serverless instances according to policies, and handle high-concurrency requests;

[0028] Storage expansion module: used to manage distributed storage clusters, automatically expand storage nodes or adjust sharding strategies based on data volume growth to ensure that the resource pool capacity meets the storage needs of large files including teaching videos;

[0029] Traffic scheduling module: used to distribute user requests through the load balancer, direct traffic to idle resource nodes based on regional strategies, and optimize cross-campus resource access latency;

[0030] Cost Optimization Module: This module intelligently mixes on-demand and spot instances to reduce computing resource costs while guaranteeing service level agreements (SLAs), accommodating budget constraints in regions with limited educational resources.

[0031] Preferably, the traffic scheduling module allocates traffic resources based on a dynamic resource allocation function, and its calculation formula is as follows:

[0032]

[0033] Where: C(t) represents the total amount of computing resources at time t; C0 represents the basic computing resource quota; L(t) represents the real-time load monitoring value; L0 represents the load benchmark threshold; γ represents the elastic scaling factor; σ represents the load sensitivity adjustment factor.

[0034] Preferably, the calculation scaling module performs scaling calculations based on the storage resource scaling model, and its calculation formula is as follows:

[0035]

[0036] Where: S(t) represents the number of storage nodes at time t; D(t) represents the total amount of real-time data; D0 represents the storage expansion trigger threshold; κ represents the storage elasticity coefficient; S min Indicates the lower limit of storage nodes; S max Indicates the upper limit of storage nodes.

[0037] Preferably, the intelligent central unit specifically includes:

[0038] Graph construction module: used to automatically build and maintain a semantic knowledge graph network that reflects the inherent logical relationships between knowledge points based on fragmented resources and metadata;

[0039] Profile building module: used to continuously analyze students' learning behavior trajectory data, build refined user models, and accurately portray their knowledge mastery, learning preferences, and potential needs;

[0040] Algorithm engine module: used to integrate and run core intelligent components including machine learning models, recommendation algorithms, and knowledge graph computing, providing analysis, matching, and prediction capabilities for the system;

[0041] Intelligent push module: used to combine user portrait features and knowledge graph semantic relationships to proactively match and recommend the most relevant teaching resources to students or teachers;

[0042] Path planning module: used to dynamically generate and optimize personalized learning path recommendations based on the user's learning goals, current knowledge status, and dependencies between knowledge points.

[0043] Preferably, the intelligent push module specifically includes:

[0044] Feature extraction module: used to analyze the metadata and content features of teaching resources, extract the learning stage, preference tags and weak point identifiers in the user profile, and generate a structured feature vector;

[0045] Intent recognition module: used to analyze the user's current behavior scenario in real time and infer immediate learning intentions based on historical interaction data;

[0046] Graph Mapping Module: This module is used to connect to the dynamic knowledge graph service, map user characteristics and resource characteristics into a unified semantic space, establish a three-layer association network of "user-knowledge point-resource", and identify the optimal matching path;

[0047] Multimodal matching module: This module is used to call the multimodal resource engine, integrate the semantic features of heterogeneous resources such as text, video, and exercises, and solve the problem of cross-modal similarity calculation;

[0048] Hybrid recommendation module: used to integrate graph neural networks and reinforcement learning models, collaboratively process static recommendations based on knowledge structure and dynamic optimization based on behavioral feedback, and generate a ranked list of candidate resources;

[0049] Scenario adaptation module: used to dynamically adjust the presentation format of pushed resources based on user terminal type, network environment, and teaching stage;

[0050] Feedback learning module: used to collect users' explicit ratings and implicit feedback on recommended resources, continuously iterate the recommendation model parameters, and achieve closed-loop optimization.

[0051] Preferably, the intelligent push module performs resource matching and accurate push based on the intelligent push priority function, and its calculation formula is as follows:

[0052]

[0053] Where: P represents the priority score of smart push; n represents the total number of summation items, i.e., the number of feature dimensions; i represents the dimension index number, i.e., the feature dimension number; S i represents the weight coefficient of resource dimension i; C i Represents the user's cognitive state value; K i Indicates resource matching degree; |C i -K i | represents the absolute value of the user's resource cognition difference; t i Indicates the time of last contact with the same type of resource; δ(t i ) represents the time-effect gain function; α represents the time-effect gain coefficient; β represents the difference sensitivity coefficient.

[0054] Preferably, the push condition of the intelligent push priority function of the intelligent push module is: P>θ-ηN, where θ is the basic decision threshold, that is, the minimum priority required for push, and the value range is 0.4≤θ≤0.7; η is the fatigue suppression coefficient, that is, the threshold attenuation of a single push, and the value range is 0.05≤η≤0.1; N is the number of pushes on the day, that is, the cumulative push count.

[0055] The technical effects and advantages of the present invention are as follows:

[0056] In the present invention, an intelligent resource allocation system is constructed through an elastic scheduling module: the load perception module collects database resource indicators in real time and predicts demand trends; the policy decision module combines a rule engine with a machine learning model, and relies on a storage resource scaling model and a dynamic resource allocation function to accurately calculate the number of storage nodes and the total amount of computing resources, so that the computing scaling module can dynamically allocate Serverless instances of the cloud database according to the policy; the storage scaling module automatically adjusts the scale of the distributed storage cluster according to the data volume; the traffic scheduling module optimizes cross-regional database access latency through load balancing and regional strategies, avoiding the frequent scheduling fluctuations of the traditional threshold method; the cost optimization module uses a mixture of on-demand instances and bidding instances to reduce the cost of cloud database services; so that the cloud database service can dynamically scale resources according to the teaching load, automatically expand resources according to the model during high concurrency periods such as teaching peaks, ensure the response speed of big data resource services, accurately reduce resource allocation during low loads, effectively avoid idle resources and waste, and at the same time improve the resource utilization of the cloud database through dynamic threshold adjustment, avoid performance bottlenecks during peak periods that lead to slow response and service interruptions, and provide elastic and efficient big data resource services and cloud database support for teaching scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The disclosure of the present invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of the present invention. In the drawings, the same reference numerals are used to refer to the same components:

[0058] Figure 1 This is a schematic diagram of the overall architecture of a cloud database-based teaching big data resource management system of the present invention;

[0059] Figure 2 A schematic diagram of a data storage process of a cloud base unit of a teaching big data resource management system based on a cloud database of the present invention;

[0060] Figure 3 This is a schematic diagram of the flexible scheduling module architecture of a cloud database-based teaching big data resource management system of the present invention;

[0061] Figure 4 This is a schematic diagram of the workflow of the flexible scheduling module of the cloud database-based teaching big data resource management system of the present invention;

[0062] Figure 5 Schematic diagram of the progressive growth of computing resources C(t) according to an S-shaped curve in a cloud database-based teaching big data resource management system of the present invention;

[0063] Figure 6 This is a schematic diagram of the intelligent push module architecture of a cloud database-based teaching big data resource management system of the present invention;

[0064] Figure 7 This is a schematic diagram of the workflow of the intelligent push module framework of the cloud database-based teaching big data resource management system of the present invention. DETAILED DESCRIPTION

[0065] It is easy to understand that according to the technical solution of the present invention, without changing the essential spirit of the present invention, a person skilled in the art can propose a variety of interchangeable structural modes and implementation modes. Therefore, the following specific embodiments and drawings are only exemplary descriptions of the technical solution of the present invention and should not be regarded as the entire invention or as a limitation or restriction of the technical solution of the present invention.

[0066] Reference Figure 1-Figure 5 As shown, the present invention provides a technical solution: a cloud database-based teaching big data resource management system. The system adopts a layered distributed architecture and consists of three core modules: a data preparation unit, a cloud base unit, and an intelligent hub unit. It is supplemented by a teaching diagnosis module and a system operation and maintenance unit to form a complete closed loop. Each module realizes data flow through a standardized API interface and a message queue. The specific description is as follows:

[0067] First, the data preparation unit: This unit is responsible for uniformly accessing, standardizing, and breaking down raw data from different teaching scenarios into manageable knowledge units, laying the foundation for subsequent processing. Specifically, it includes:

[0068] Multi-source access module: responsible for connecting and integrating heterogeneous data sources (structured / unstructured) from various teaching scenarios (online courses, classroom recordings, homework, examination systems, etc.) to achieve unified access and preliminary cleaning of data.

[0069] Metadata management module: responsible for defining and maintaining unified resource metadata standards (type, subject, authority, etc.), ensuring that all stored resources have standardized, complete and interoperable descriptive information to support efficient retrieval and sharing.

[0070] Resource fragmentation module: Use natural language processing (NLP) and machine learning technology to intelligently split teaching content (blackboard writing, PPT, voice, etc.) to form fine-grained knowledge slices, and establish an association network between slices and knowledge points and user behavior.

[0071] Second, the cloud base unit: This unit provides elastically scalable cloud storage architecture and infrastructure based on cloud databases, ensuring elastic resource expansion, high availability, security, reliability, and cross-regional collaboration. Specifically, it provides:

[0072] (1) Resource pool module: It provides a standardized, high-performance storage engine based on cloud database to centrally store and manage all standardized teaching resources (raw data, fragmented resources, metadata) to form a unified data pool.

[0073] (2) Elastic Scheduling Module: Dynamically allocates cloud database storage space, computing resources (CPU / memory), and network bandwidth based on real-time access pressure (such as peak hours during exam weeks) to ensure stable system performance and optimize resource utilization. Specifically, it includes:

[0074] Load sensing module: collects CPU, memory, network bandwidth, and storage IOPS indicators of each node in the teaching system in real time, and predicts resource demand trends based on business scenarios (such as exam registration peaks) to provide data support for dynamic scheduling.

[0075] Policy decision module: Built-in rule engine and machine learning models (such as time series prediction) formulate resource allocation strategies based on load data, such as automatically triggering the expansion of Serverless computing units during the opening period of elective courses.

[0076] Computing and scaling module: Connects to serverless computing services (such as AWS Lambda / AliyunFC), dynamically creates or releases serverless instances according to policies, and handles high-concurrency requests (such as the traffic impact of tens of thousands of course registrations).

[0077] The computing scaling module performs scaling calculations based on the storage resource scaling model. The calculation formula is as follows:

[0078]

[0079] Where: S(t) represents the number of storage nodes at time t; D(t) represents the total amount of real-time data; D0 represents the storage expansion trigger threshold; κ represents the storage elasticity coefficient; S min Indicates the lower limit of storage nodes; S max Indicates the upper limit of storage nodes.

[0080] Storage Scaling Module: Manages distributed storage clusters (Ceph / HDFS), automatically expands storage nodes or adjusts sharding strategies based on data growth, ensuring that the resource pool capacity meets the storage needs of large files such as teaching videos.

[0081] Traffic scheduling module: distributes user requests through the load balancer, and directs traffic to idle resource nodes in combination with regional strategies (such as proximity access), thereby optimizing cross-campus resource access latency.

[0082] The traffic scheduling module allocates traffic resources based on the dynamic resource allocation function. The calculation formula is as follows:

[0083]

[0084] In the formula: C(t) represents the total amount of computing resources at time t; C0 represents the basic computing resource quota; L(t) represents the real-time load monitoring value (request per second); L0 represents the load reference threshold; γ represents the elastic expansion coefficient (0.5 ≤ γ ≤ 2); σ represents the load sensitivity adjustment factor.

[0085] Resource dynamic allocation rules:

[0086] Low-load steady state

[0087] When the real-time load L(t) is lower than the threshold L0 (such as concurrent requests <5000), the system maintains the basic resource configuration C0 (such as 100 container instances), avoiding resource waste and ensuring the optimal baseline service cost;

[0088] High-load elastic expansion

[0089] When L(t) exceeds the threshold L0 (such as during the peak period of exam registration), the computing resources C(t) gradually increase according to an S-shaped curve:

[0090] Initial gentle segment: Prevents triggering false expansion due to slight fluctuations;

[0091] Rapidly rising segment: Responds to load surges;

[0092] Saturation segment in later period: Avoids over-provisioning.

[0093] The hyperbolic tangent function tanh is used to achieve smooth scaling of the expansion / contraction, eliminating the frequent resource scaling (such as ±30% instance number fluctuation within 5 minutes) caused by traditional threshold methods.

[0094] Cost optimization module: Intelligent hybrid use of on-demand instances and SpotInstance (bid for instances) to reduce computing resource costs while ensuring service SLA, and adapt to budget constraints in education resource scarce areas.

[0095] (3) Cross-domain synchronization module: Utilize the multi-region deployment capability of cloud databases to achieve real-time or near-real-time data synchronization of teaching resources between different campuses and cooperative institutions, providing a consistent data view for cross-regional collaboration.

[0096] (4) Encryption storage module: Apply strong encryption technology in data transmission (such as TLS / SSL) and static storage (such as AES-256) to ensure the confidentiality and security of teaching resources (especially sensitive data) when stored in the cloud.

[0097] (5) Disaster recovery module: Based on the distributed storage characteristics of cloud databases, automatically implement multi-copy storage and regular incremental / full backup strategies to ensure data persistence, high availability, and disaster recovery capabilities.

[0098] Third, the intelligent central unit: used to conduct in-depth analysis and mining of teaching data through the use of knowledge graphs, user portraits and machine learning algorithms to drive intelligent services.

[0099] Fourth, the Teaching Application Unit: This unit is used to convert data analysis results into diagnostic insights and optimization suggestions that directly serve teaching practice. Specifically:

[0100] Teaching diagnosis module: used to aggregate and analyze multi-dimensional data (interaction rate, homework quality, test scores, etc.) during the teaching process, providing teachers with objective teaching effect evaluation and difficulty identification;

[0101] Decision support module: Aggregates data from the entire teaching process (classroom interaction rate, homework completion quality, test scores, etc.), generates a visual education quality assessment report through a multi-dimensional analysis model, and provides managers with resource allocation optimization suggestions and teaching quality improvement directions to support accurate decision-making.

[0102] Open platform module: Encapsulates the teaching resource service interface through a standardized API gateway, allowing third-party applications to securely call question banks, courseware, videos and other resources after OAuth2.0 authentication, realizing cross-platform resource sharing and function integration, and building an open education ecosystem.

[0103] System Operation and Maintenance Unit: This unit is used to ensure the safe and compliant operation of the entire system, authority control, and continuous performance monitoring. It is divided into:

[0104] Permission management module: used to implement refined role-based access control (RBAC), strictly defining the operation permissions of different roles such as students, teachers, and administrators for various teaching resources (such as grades and lesson plans), ensuring data privacy and compliance;

[0105] System monitoring module: used to collect and visualize key system indicators (resource utilization, performance, access volume, exception logs) in real time, providing data support for resource elastic scheduling, troubleshooting and operation and maintenance decision-making.

[0106] Specifically, the data preparation unit integrates and cleans heterogeneous data from multiple scenarios through the multi-source access module, the metadata management module defines a unified standard, and the resource fragmentation module uses technology to split the teaching content into knowledge slices and establish associations; the resource pool module of the cloud base unit centrally stores resources, the load perception module in the elastic scheduling module collects indicators and predicts demand, the policy decision module formulates strategies based on the model, the computing scaling module dynamically adjusts the number of storage nodes based on the relationship between the total amount of real-time data and the trigger threshold based on the storage resource scaling model, and adjusts computing resources based on the comparison of real-time load and benchmark threshold based on the dynamic resource allocation function, the storage scaling module manages the cluster to achieve storage expansion and sharding adjustment, the traffic scheduling module distributes requests through load balancing and regional strategies, the cost optimization module uses mixed instances, and the cross-domain synchronization module The blocks realize data synchronization, and the encryption and disaster recovery modules ensure security; the intelligent central unit uses technology to analyze data, the teaching diagnosis module aggregates data to provide evaluation, the decision support module generates reports, the open platform module realizes sharing through API, the permission management module of the system operation and maintenance unit implements access control, and the monitoring module collects indicators; the system can dynamically adjust the cloud database storage capacity and computing power according to the teaching load, automatically expand the capacity to cope with traffic during peak hours, maintain the basic configuration during low load to avoid waste, eliminate the frequent scaling problems caused by traditional methods, combine load balancing and regional strategies to optimize latency, and reduce costs through hybrid instances, while ensuring data synchronization, encryption and disaster recovery, realizing elastic resource expansion, high availability, security and reliability, providing efficient support for teaching diagnosis, decision-making and open ecology, and solving the problems of resource waste and performance bottlenecks.

[0107] Reference Figure 1 、 Figure 6 and Figure 7 As shown, in this embodiment: the intelligent central unit specifically includes:

[0108] (1) Graph construction module: Based on fragmented resources and metadata, it automatically constructs or maintains a teaching knowledge graph that reflects the relationship between knowledge points, providing a semantic basis for intelligent recommendation and resource association.

[0109] (2) Portrait construction module: Continuously analyze students' learning behavior data (click stream, duration, grades, interactions, etc.) to build a detailed user portrait that describes their knowledge level, learning style preferences, and potential needs.

[0110] (3) Algorithm engine module: Integrates and runs core computing logic such as machine learning models, recommendation algorithms, and knowledge graph query engines to provide intelligent driving for resource matching, path planning, and teaching diagnosis.

[0111] (4) Intelligent push module: Using the algorithm engine, combined with user portraits and knowledge graphs, it proactively matches and pushes the most relevant teaching resources (such as micro-courses and exercises targeting weak points) to students or teachers; specifically, it includes:

[0112] Feature extraction module: Analyzes the metadata and content features of teaching resources (such as knowledge point labels, difficulty coefficients, and media types), and extracts the learning stages, preference labels, and weakness identifiers in user portraits to generate structured feature vectors.

[0113] Intent Recognition Module: Analyzes the user's current behavior scenario in real time (such as repeatedly watching a video clip or requesting to re-do a wrong question), and combines historical interaction data to infer immediate learning intentions (such as concept reinforcement, skill development, or test review).

[0114] Graph mapping module: connects to the dynamic knowledge graph service, maps user characteristics and resource characteristics into a unified semantic space, establishes a three-layer association network of "user-knowledge point-resource", and identifies the optimal matching path.

[0115] Multimodal matching module: Calls the multimodal resource engine (MM-Engine) to integrate the semantic features of heterogeneous resources such as text, video, and exercises to solve the problem of cross-modal similarity calculation (such as matching blackboard content with explanation videos).

[0116] Hybrid recommendation module: Integrates graph neural network (GNN) and reinforcement learning model to collaboratively process static recommendation based on knowledge structure and dynamic optimization based on behavioral feedback to generate a ranked list of candidate resources.

[0117] Scenario adaptation module: Dynamically adjusts the presentation format of pushed resources (such as converting long videos to micro-class clips) based on the user terminal type (PC / mobile), network environment, and teaching stage (pre-class preparation / post-class consolidation).

[0118] Feedback learning module: collects users' explicit ratings (such as likes / collections) and implicit feedback (such as dwell time / skip rate) on recommended resources, continuously iterates the recommendation model parameters, and achieves closed-loop optimization.

[0119] The intelligent push module performs resource matching and precise push based on the intelligent push priority function. The calculation formula is as follows:

[0120]

[0121] Where: P represents the priority score of smart push; n represents the total number of summation items, i.e., the number of feature dimensions (fixed to 3); i represents the dimension index number, i.e., the feature dimension number (i=1: knowledge point, i=2: difficulty, i=3: media type); S irepresents the weight coefficient of resource dimension i {S1 = 0.6 (knowledge point weight), S2 = 0.3 (difficulty weight), S3 = 0.1 (media type weight)}; C i Indicates the user's cognitive state value (0≤C i ≤10, indicating the user's ability level in the current dimension); K i Indicates resource matching degree (0≤K i ≤1, indicating the correlation between resources and target dimensions); |C i -K i | represents the absolute value of the user's resource cognition difference; t i Indicates the time of last contact with the same type of resource (t i ≥0, indicating the length of the time interval); δ(t i ) represents the time-effectiveness gain function, {0≤δ≤1, δ=min(1,ti / 7), indicating linear growth to saturation after 7 days}; α represents the time-effectiveness gain coefficient (0.1≤α≤0.3, indicating the intensity of the influence of timeliness on priority); β represents the difference sensitivity coefficient (0.3≤β≤0.7, indicating the intensity of the penalty for controlling cognitive differences).

[0122] The push condition of the intelligent push priority function of the intelligent push module is: P>θ-ηN, where θ is the basic decision threshold, that is, the minimum priority required for push, and its value range is 0.4≤θ≤0.7; η is the fatigue suppression coefficient, that is, the threshold attenuation of a single push, and its value range is 0.05≤η≤0.1; N is the number of pushes on the day, that is, the cumulative push count.

[0123] (5) Path planning module: Based on user profiles (learning goals, mastery level) and knowledge graphs (knowledge point dependencies), it dynamically generates and optimizes personalized learning path recommendations and resource push sequences.

[0124] Specifically, the graph construction module automatically constructs and maintains a teaching knowledge graph that reflects the relationship between knowledge points based on fragmented resources and metadata, laying a semantic foundation for intelligent recommendation; the portrait construction module continuously analyzes students' learning behavior data, including click streams, duration, grades, etc., to form a refined user portrait that depicts the level of knowledge mastery and learning style preferences; the algorithm engine module integrates core computing logic such as machine learning models and recommendation algorithms to provide intelligent drive for resource matching, teaching diagnosis, etc. As a key link, the intelligent push module uses the feature extraction module to parse resource metadata and user portraits to generate feature vectors, uses the intent recognition module to infer learning intentions based on behavioral scenarios, and uses the graph mapping module to establish a "user-knowledge point-resource" association network in a unified semantic space. It then uses multimodal matching technology to integrate the semantics of heterogeneous resources such as text and video, and combines the hybrid recommendation model composed of graph neural network and reinforcement learning to generate a list of candidate resources. At the same time, it introduces the time gain function and fatigue suppression mechanism to calculate the push priority and achieve accurate push according to conditions; the path planning module dynamically generates and optimizes personalized learning paths and resource push sequences based on the knowledge point dependency relationship and user portrait of the knowledge graph; through the dynamic association of the knowledge graph and the precise capture of user needs, it realizes the intelligent matching of teaching resources, improves the accuracy of resource push through multimodal fusion and recommendation algorithm optimization, avoids information overload, and enhances the targeted learning through personalized learning path generation, effectively solving problems such as low resource matching efficiency and insufficient learning adaptability, promoting the transformation of teaching services towards intelligence and personalization, and improving the learning experience and the efficiency of teaching resource utilization.

[0125] The technical scope of the present invention is not limited to the contents of the above description. Those skilled in the art can make various deformations and modifications to the above embodiments without departing from the technical idea of ​​the present invention, and these deformations and modifications should all fall within the protection scope of the present invention.

Claims

1. A teaching big data resource management system based on cloud database, characterized in that: include: Data preparation unit: responsible for unified access, standardized description, and decomposition of raw data from different teaching scenarios into manageable knowledge units, laying the foundation for subsequent processing; Cloud base unit: used to provide elastically scalable cloud storage architecture and infrastructure support, specifically: Resource pool module: used to provide centralized and standardized cloud storage services, centrally manage all prepared teaching resource data, and form an efficient access data pool; Elastic Scheduling Module: This module dynamically allocates and adjusts the computing power, storage space, and network bandwidth resources of the cloud database based on real-time access load to ensure stable system performance. Cross-region synchronization module: used to achieve real-time or near-real-time data synchronization of teaching resources between nodes in different regions of the cloud, and implement cross-campus or cross-institutional educational resource sharing and collaboration; Encrypted storage module: used to apply encryption technology on data transmission links and static storage media to protect the confidentiality of sensitive data in teaching resources in the cloud; and Disaster recovery module: This module implements redundant storage of multiple copies of data through a distributed storage architecture and executes regular backup strategies to ensure data persistence and disaster recovery capabilities. Intelligent central unit: used to conduct in-depth analysis and mining of teaching data through the use of knowledge graphs, user portraits and machine learning algorithms to drive intelligent services.

2. The cloud database-based teaching big data resource management system according to claim 1, characterized in that: The system also includes a teaching diagnosis module and a system operation and maintenance unit: Teaching Diagnosis Module: This module is used to comprehensively analyze multi-dimensional information including classroom interaction data, homework completion, and test scores to evaluate the overall teaching effect and identify key teaching difficulties. System Operation and Maintenance Unit: This unit is used to ensure the safe and compliant operation of the entire system, authority control, and continuous performance monitoring. It is divided into: Permission management module: used to implement role-based access control strategies, define the operation permissions of different user roles for various teaching resources, and ensure data security and privacy protection; System monitoring module: used to collect, analyze and visualize the system's key performance indicators, resource usage status and operation logs in real time, providing a decision-making basis for system operation and maintenance, optimization and troubleshooting.

3. The cloud database-based teaching big data resource management system according to claim 1, characterized in that: The data preparation unit includes: Multi-source access module: This module is responsible for connecting and aggregating structured and unstructured data sources generated by teaching scenarios, including online courses, classroom recordings, homework systems, and examination platforms, to achieve unified data entry and preliminary cleaning. Metadata management module: responsible for formulating, maintaining and implementing unified resource description metadata standards to ensure that all teaching resources have complete, consistent and interoperable metadata information; Resource fragmentation module: used to use natural language processing and machine learning technologies to intelligently break down complex teaching content into fine-grained knowledge slices or units, and establish their association with knowledge points and user behavior.

4. The cloud database-based teaching big data resource management system according to claim 1, characterized in that: The flexible scheduling module specifically includes: Load sensing module: used to collect CPU, memory, network bandwidth, and storage IOPS indicators of each node in the teaching system in real time, predict resource demand trends based on business scenarios, and provide data support for dynamic scheduling; Policy decision module: Used for the built-in rule engine and machine learning model to formulate resource allocation strategies based on load data; Compute scaling module: used to connect to the Serverless computing service, dynamically create or release serverless instances according to policies, and handle high-concurrency requests; Storage expansion module: used to manage distributed storage clusters, automatically expand storage nodes or adjust sharding strategies based on data volume growth to ensure that the resource pool capacity meets the storage needs of large files including teaching videos; Traffic scheduling module: used to distribute user requests through the load balancer, direct traffic to idle resource nodes based on regional strategies, and optimize cross-campus resource access latency; Cost Optimization Module: This module intelligently mixes on-demand and spot instances to reduce computing resource costs while guaranteeing service level agreements (SLAs), accommodating budget constraints in regions with limited educational resources.

5. The cloud database-based teaching big data resource management system according to claim 4, characterized in that: The traffic scheduling module allocates traffic resources based on the dynamic resource allocation function, and its calculation formula is as follows: Where: C(t) represents the total amount of computing resources at time t; C0 represents the basic computing resource quota; L(t) represents the real-time load monitoring value; L0 represents the load benchmark threshold; γ represents the elastic scaling factor; σ represents the load sensitivity adjustment factor.

6. The cloud database-based teaching big data resource management system according to claim 4, characterized in that: The calculation and scaling module performs scaling calculations based on the storage resource scaling model, and its calculation formula is as follows: Where: S(t) represents the number of storage nodes at time t; D(t) represents the total amount of real-time data; D0 represents the storage expansion trigger threshold; κ represents the storage elastic coefficient; S min Indicates the lower limit of storage nodes; S max Indicates the upper limit of storage nodes.

7. The cloud database-based teaching big data resource management system according to claim 1, characterized in that: The intelligent central unit specifically includes: Graph construction module: used to automatically build and maintain a semantic knowledge graph network that reflects the inherent logical relationships between knowledge points based on fragmented resources and metadata; Profile building module: used to continuously analyze students' learning behavior trajectory data, build refined user models, and accurately portray their knowledge mastery, learning preferences, and potential needs; Algorithm engine module: used to integrate and run core intelligent components including machine learning models, recommendation algorithms, and knowledge graph computing, providing analysis, matching, and prediction capabilities for the system; Intelligent push module: used to combine user portrait features and knowledge graph semantic relationships to proactively match and recommend the most relevant teaching resources to students or teachers; Path planning module: used to dynamically generate and optimize personalized learning path recommendations based on the user's learning goals, current knowledge status, and dependencies between knowledge points.

8. The cloud database-based teaching big data resource management system according to claim 7, characterized in that: The intelligent push module specifically includes: Feature extraction module: used to analyze the metadata and content features of teaching resources, extract the learning stage, preference tags and weak point identifiers in the user profile, and generate a structured feature vector; Intent recognition module: used to analyze the user's current behavior scenario in real time and infer immediate learning intentions based on historical interaction data; Graph Mapping Module: This module connects to the dynamic knowledge graph service, maps user and resource features into a unified semantic space, establishes a three-layer association network of "user-knowledge point-resource", and identifies the optimal matching path. Multimodal matching module: This module is used to call the multimodal resource engine, integrate the semantic features of heterogeneous resources such as text, video, and exercises, and solve the problem of cross-modal similarity calculation; Hybrid recommendation module: used to integrate graph neural networks and reinforcement learning models, collaboratively process static recommendations based on knowledge structure and dynamic optimization based on behavioral feedback, and generate a ranked list of candidate resources; Scenario adaptation module: used to dynamically adjust the presentation format of pushed resources based on user terminal type, network environment, and teaching stage; Feedback learning module: used to collect users' explicit ratings and implicit feedback on recommended resources, continuously iterate the recommendation model parameters, and achieve closed-loop optimization.

9. The cloud database-based teaching big data resource management system according to claim 8, characterized in that: The intelligent push module performs resource matching and accurate push based on the intelligent push priority function, and its calculation formula is as follows: Where: P represents the smart push priority score; n represents the total number of summed items, i.e., the number of feature dimensions; i represents the dimension index number, i.e., the feature dimension number; S i represents the weight coefficient of resource dimension i; C i Represents the user's cognitive state value; K i Indicates resource matching degree; |C i -K i | represents the absolute value of the user's resource cognition difference; t i Indicates the time of last contact with the same type of resource; δ(t i ) represents the time-effect gain function; α represents the time-effect gain coefficient; β represents the difference sensitivity coefficient.

10. The cloud database-based teaching big data resource management system according to claim 9, characterized in that: The push condition of the intelligent push priority function of the intelligent push module is: P>θ-ηN, where θ is the basic decision threshold, that is, the minimum priority required for push, and the value range is 0.4≤θ≤0.7; η is the fatigue suppression coefficient, that is, the threshold attenuation of a single push, and the value range is 0.05≤η≤0.1; N is the number of pushes on the day, that is, the cumulative push count.

Images