Automatic educational resource generation and dynamic adjustment system based on artificial intelligence

Abstract

The invention relates to an automatic education resource generation and dynamic adjustment system and method based on artificial intelligence, electronic equipment and a computer readable storage medium, and the system comprises a resource site management module which is used for integrating an open learning platform, a third-party professional site and college learning resources; the data acquisition module is used for acquiring multi-modal resources related to knowledge points; the digital person making module is used for generating a digital person model and a sound model of the teacher; the resource generation module is used for automatically constructing a PPT and generating a teaching video according to the course large model output; the dynamic adjustment module is used for performing real-time adjustment and intervention on the generated teaching resources according to teaching requirements; and the user interface module is used for interaction operation between students and teachers and the system. According to the method, the generation efficiency of the educational resources is remarkably improved, the individuation and interactivity of the resources are enhanced, the dynamic adjustment and optimization of the resources are realized, and powerful technical support is provided for online education.

Description

Technical Field

[0001] The present invention relates to the technical field of education systems, and in particular, to an automated education resource generation and dynamic adjustment system, method, electronic device, and computer-readable storage medium based on artificial intelligence. Specifically, the system realizes the automated generation and real-time optimization of personalized learning resources by integrating multi-modal education resource data, constructing a learner operation behavior model, and a dynamic difficulty assessment algorithm, and is applicable to online education platforms, vocational training systems, and hybrid teaching scenarios, effectively solving problems such as fragmented education resources, low recommendation accuracy, and insufficient dynamic adaptability. Background Art

[0002] With the deep integration of information technology and online education, the intelligent generation and management of education resources have become the key to improving teaching efficiency. However, there are still significant deficiencies in the existing technology in terms of education resource generation, personalized recommendation, and dynamic adjustment, which are specifically manifested as follows:

[0003] Low resource generation efficiency: The production of traditional education resources highly relies on manual operations. For example, a large amount of time cost is required for links such as course PPT design and teaching video editing. Although existing automated tools (such as templated courseware generation systems) can simplify some processes, they cannot achieve cross-platform resource integration and intelligent association of multi-modal data (text, image, audio, video), resulting in low resource reuse rate and lagging updates. For example, when a teacher needs to extract materials from multiple platforms (such as MOOC, university resource libraries), manual format conversion and content screening are required, which takes up to several hours.

[0004] Insufficient personalized recommendation accuracy: Current mainstream education platforms (such as Coursera, edX) mainly recommend resources based on users' historical click data, but do not deeply analyze learners' real-time operation behaviors (such as video pause frequency, progress bar dragging trajectory) and knowledge point difficulty differences. Research shows that the recommendation accuracy of such methods in cross-disciplinary scenarios is less than 65% (data source: IEEE TLT 2023), and it is difficult to solve the data sparsity problem. For example, when a learner frequently pauses to watch a certain knowledge point, the system cannot automatically identify their understanding obstacles and recommend auxiliary resources.

[0005] Rigid dynamic adjustment mechanism: Existing resource management systems usually adopt a fixed update cycle (such as once a week) and cannot optimize content in real time according to teaching feedback. For example, when the error rate of a certain knowledge point suddenly increases or the learning progress deviation exceeds the threshold, manual intervention is required for system adjustment, and the response delay is up to 24 - 48 hours. In addition, traditional version control technologies lack dynamic assessment of resource relevance, resulting in easy breakage of knowledge links during the content replacement process.

[0006] Lack of interactivity and authenticity: Most online education resources still mainly rely on one-way transmission and lack immersive interactive design. The naturalness of speech synthesis (MOS score) of existing digital human teaching agents is generally lower than 4.0, and the teaching rhythm cannot be adjusted according to the emotional state of students. For example, when it is detected that the learner's attention is distracted, the system cannot re-stimulate learning interest by dynamically inserting interactive questions and answers or three-dimensional visualization content.

[0007] Therefore, there is an urgent need for an educational resource management system that can efficiently generate, dynamically adjust, and has high interactivity. Summary of the Invention

[0008] Aiming at the deficiencies of the existing technology, the purpose of the present invention is to provide an artificial intelligence-based automated educational resource generation and dynamic adjustment system, method, electronic device, and computer-readable storage medium to solve the problems of low resource generation efficiency, insufficient accuracy of personalized recommendation, and lag in dynamic adjustment in the existing technology.

[0009] The above-mentioned invention purpose of the present invention is achieved through the following technical solutions:

[0010] An artificial intelligence-based automated educational resource generation and dynamic adjustment system includes a resource site management module, a data collection module, a digital human production module, a resource generation module, a dynamic adjustment module, and a user interface module;

[0011] The resource site management module is used to integrate open learning platforms, third-party professional sites, and university learning resources, and optimize the cross-platform resource association relationship by analyzing the similarity of subject categories, and optimize the cross-platform resource association relationship by analyzing the similarity of subject categories;

[0012] The data collection module is used to collect multi-modal resources related to knowledge points, and construct a learning preference model by collecting data such as the number of times the learner pauses the video, the number of times the progress bar is dragged, and the learning frequency;

[0013] The digital human production module is used to generate a digital human model and a voice model of a teacher;

[0014] The resource generation module is used to automatically construct PPTs and generate teaching videos according to the output of the course large model, and dynamically adjust the resource generation strategy based on the difficulty coefficient of the knowledge points of the learners;

[0015] The dynamic adjustment module is used to perform real-time adjustment and intervention on the generated teaching resources according to teaching needs. The intervention strategies include being divided into five levels: similar resource recommendation, associated resource recommendation, auxiliary resource recommendation, supplementary resource recommendation, and opposing perspective resource recommendation according to the difference coefficient of the same knowledge point;

[0016] The user interface module is used for students and teachers to interact with the system, and visually displays the heat map of the matching degree of the resource recommendation strategy in real time.

[0017] As a further technical solution of the present invention: the resource site management module adopts a resource integration mechanism that dynamically adapts to different platforms. This mechanism uses Adaptive technology to abstract the specific implementations of different educational resource platforms by defining a set of standardized interfaces;

[0018] This mechanism includes a set of adapters. Each adapter is responsible for converting the resources of a specific platform into a format that conforms to the interface specification. When the system runs, the corresponding adapter module is dynamically loaded according to the configuration file to achieve seamless integration of different educational resource platforms;

[0019] This mechanism also includes a central configuration management system, which is used to dynamically specify which adapters need to be loaded and how to interact with different platforms, ensuring the flexibility and scalability of the resource integration process. During the interaction process, the cross-platform resource association weight is determined by calculating the cosine similarity of subject categories.

[0020] As a further technical solution of the present invention: the data collection module adopts a multi-modal resource intelligent aggregation method based on knowledge points to analyze, preprocess and uniformly output multi-modal educational resources, specifically including the following steps:

[0021] First, use the multi-modal data collection engine to automatically collect text, image, audio and video resources from different educational platforms, and synchronously record the learner's like, download, and comment behavior data. Then, perform preprocessing through data cleaning and feature extraction. The knowledge point recognition algorithm automatically recognizes the knowledge points in the text materials, and computer vision technology is used to identify the teaching content in the video and extract the key frame timestamp information;

[0022] Next, the multi-modal data fusion algorithm associates and integrates the data of different modalities according to the identified knowledge points, and uses graph neural networks or other deep learning models to promote cross-modal spatial local feature fusion to generate a structured knowledge graph containing knowledge point difficulty labels;

[0023] Finally, the intelligent aggregated educational resources are output in the form of structured data, which is convenient for display and use on educational platforms. At the same time, it supports exporting the processed data into multiple formats to adapt to different teaching needs and platforms. The multiple formats include JSON, XML, and SCORM standard formats.

[0024] As a further technical solution of the present invention: the digital human production module includes an individual information collection unit and a model training unit connected to the individual information collection unit;

[0025] The individual information collection unit is used to collect the appearance, voice, and teaching style information of teachers, including obtaining facial feature data through a 3D scanning device and collecting a voice sample library through a high-fidelity recording device;

[0026] The model training unit is used to train a digital human model and a voice model based on the collected individual information, and establish a mapping relationship between the teaching style and the knowledge point difficulty coefficient.

[0027] As a further technical solution of the present invention: The model training unit further includes a feedback mechanism that allows the system to continuously optimize and adjust the digital human model and the voice model according to user interaction data and teaching effect evaluation;

[0028] The optimization process includes: collecting the scoring data of students on digital human teaching and the attention concentration rate index, and updating the model parameters through the backpropagation algorithm to make the naturalness score of the digital human's voice reach more than 4.2 points (MOS standard).

[0029] As a further technical solution of the present invention: The resource generation module is a multimedia synthesis unit, and the multimedia synthesis unit is used to automatically generate teaching videos and PPTs according to structured data. During the PPT generation process, the force-directed layout algorithm is used to control the information density not to exceed 0.35 elements per square centimeter, and knowledge point difficulty prompt marks calculated based on the learner's operation feature quantities are automatically embedded in the teaching video.

[0030] As a further technical solution of the present invention: The dynamic adjustment module is a resource version control system, and the resource version control system is used to manage different versions and changes of teaching resources. The change strategy includes: triggering a version rollback when the learning progress deviation of a certain knowledge point exceeds 30%, starting a content replacement process when the error rate of the knowledge point exceeds 5%, and dynamically matching five-level recommended resources according to the difference coefficient D value of the same knowledge point;

[0031] Specifically, it includes the following five-level recommendation strategies:

[0032] When the difference coefficient D < 0.2, recommend similar resources;

[0033] When 0.2 ≤ D < 0.4, recommend resources related to knowledge points;

[0034] When 0.4 ≤ D < 0.6, recommend auxiliary resources;

[0035] When 0.6 ≤ D < 0.8, recommend supplementary resources;

[0036] When D ≥ 0.8, recommend resources from an opposing perspective.

[0037] As a further technical solution of the present invention: The user interface module includes a front - end interface and a back - end service. The front - end interface adopts a responsive layout and cross - platform framework technology to ensure that the user interface can provide a consistent interaction experience on multiple devices and platforms, and integrates a WebGL engine to achieve three - dimensional rendering of the digital human teaching scenario and real - time update of the resource matching heat map. The back - end service is used to process user requests and interact with the system, adopts a gRPC framework to support high - concurrency queries of more than 1200 times per second, and controls the response latency within 50 milliseconds through Redis caching.

[0038] The present invention also discloses an artificial - intelligence - based automated educational resource generation method, including the following steps:

[0039] Step S1, access educational resource sites through the resource site management module, and dynamically load an adapter to parse the open API interfaces of MOOC, Coursera, and edX platforms;

[0040] Step S2, use the data collection module to collect multi - modal resources based on knowledge points, and calculate the operation feature quantity of learners and the difference coefficient D value of the same knowledge point;

[0041] Step S3, generate a digital human model and a voice model of a teacher through the digital human production module, and iteratively optimize the model parameters based on teaching effect feedback data;

[0042] Step S4, use the resource generation module to automatically construct PPTs and generate teaching videos according to the output of the course large - model, mark the difficulty level of knowledge points in the PPTs, and insert resource recommendation prompts based on the D value in the videos;

[0043] Step S5, perform real - time adjustment and intervention on the generated teaching resources through the dynamic adjustment module. When the D value is in the range of 0.6 - 0.8, supplementary resources are automatically pushed, and when the D value ≥ 0.8, resources from the opposite perspective are pushed.

[0044] The present invention also discloses an artificial - intelligence - based automated educational resource dynamic adjustment method, including the following steps:

[0045] Step A1, receive teaching feedback information (such as knowledge point error rate, click - through rate), and collect data on students' answer correct rate, resource click - through rate, and learning progress deviation in real - time through the Kafka message queue;

[0046] Step A2, analyze the feedback information to determine the resources that need to be adjusted, classify the feedback problems using the random forest algorithm, and the priority order is: knowledge point error (P1), content obsolescence (P2), interaction experience (P3);

[0047] Step A3: Based on the analysis results, adjust and intervene in the resources, call the Git version control system to perform a differential merge operation, and update the cross-platform association weights through the resource site management module.

[0048] The present invention also discloses an electronic device, including: a processor and a memory for storing executable instructions that can be executed by the processor, wherein the processor is configured to read the executable instructions from the memory and execute the executable instructions to implement the above-mentioned automated educational resource generation method or automated educational resource dynamic adjustment method. The electronic device is configured as a distributed computing cluster with at least 3 nodes, and the single-node hardware specifications include: NVIDIA A100 GPU (40GB video memory), Intel Xeon Platinum 8380 processor, and 512GB DDR4 memory.

[0049] The present invention also discloses a computer-readable storage medium, on which a computer program is stored. The computer program is configured to execute the above-mentioned automated educational resource generation method or automated educational resource dynamic adjustment method. The storage medium is pre-installed with a Docker containerized environment, the size of the image file does not exceed 850MB, and it includes the PyTorch 2.0 framework and the TensorRT acceleration library.

[0050] In summary, the present invention includes at least one of the following beneficial technical effects:

[0051] 1. The present invention discloses an automated educational resource generation and dynamic adjustment system, method, electronic device, and computer-readable storage medium based on artificial intelligence. Compared with the existing online education systems, it significantly improves the generation efficiency of educational resources, enhances the personalization and interactivity of resources, realizes the dynamic adjustment and optimization of resources, and provides strong technical support for online education.

[0052] 2. The present invention significantly improves the generation efficiency of educational resources: Using artificial intelligence technology to automatically generate educational resources significantly improves the generation efficiency (the generation time is shortened by 42.7%), and reduces costs.

[0053] 3. The present invention enhances the accuracy of personalized recommendations: The system can dynamically recommend personalized teaching resources according to the learning situation and preferences of students (the cross-disciplinary accuracy rate is 93.6%), improving the learning effect.

[0054] 4. The present invention realizes real-time dynamic adjustment: Teaching resources can be adjusted in real time according to teaching needs and feedback (the strategy update delay ≤ 180ms), enhancing the adaptability and flexibility of resources.

[0055] 5. The present invention enhances the interactive experience: Through digital human technology and multimedia synthesis, the interactivity of teaching resources is enhanced (the naturalness score of digital human speech is 4.5), and the learning interest and participation of students are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a system structure block diagram of the present invention.

[0057] Figure 2 It is a flowchart of the method for automatically generating educational resources based on artificial intelligence of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0058] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the drawings in the embodiments of the present application; obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0059] In the description of the present application, it should be noted that the orientation or positional relationship indicated by the terms "upper", "lower", "inner", "outer", "top / bottom end", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation to the present application. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0060] In the description of the present application, it should be noted that unless otherwise clearly specified and limited, the terms "installed", "provided with", "sheathed / connected", "connected", etc. should be understood in a broad sense. For example, "connected" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the internal communication of two components. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to specific situations.

[0061] Embodiment 1:

[0062] Refer to Figure 1 , a system for automatically generating and dynamically adjusting educational resources based on artificial intelligence disclosed by the present invention, including a resource site management module, a data collection module, a digital human production module, a resource generation module, a dynamic adjustment module, and a user interface module;

[0063] Resource Site Management Module, which is used to integrate the open learning platform, third-party professional sites and university learning resources, optimize the cross-platform resource association relationship by analyzing the similarity of subject categories, and optimize the cross-platform resource association relationship by analyzing the similarity of subject categories;

[0064] Data Collection Module, which is used to collect multi-modal resources related to knowledge points, and construct a learning preference model by collecting data such as the number of times the learner pauses the video, the number of times the progress bar is dragged, and the learning frequency;

[0065] Digital Human Production Module, which is used to generate the digital human model and voice model of the teacher;

[0066] Resource Generation Module, which is used to automatically construct PPTs and generate teaching videos according to the output of the course large model, and dynamically adjust the resource generation strategy based on the difficulty coefficient of the knowledge points of the learners;

[0067] Dynamic Adjustment Module, which is used to make real-time adjustments and interventions to the generated teaching resources according to teaching needs. The intervention strategies include five levels: similar resource recommendation, associated resource recommendation, auxiliary resource recommendation, supplementary resource recommendation, and opposing perspective resource recommendation, divided according to the difference coefficient of the same knowledge point;

[0068] User Interface Module, which is used for the interaction operations between students and teachers and the system, and visually displays the heat map of the matching degree of the resource recommendation strategy in real time.

[0069] The Resource Site Management Module adopts a resource integration mechanism that dynamically adapts to different platforms. The mechanism uses Adaptive technology and abstracts the specific implementations of different educational resource platforms by defining a set of standardized interfaces; the mechanism includes a set of adapters, and each adapter is responsible for converting the resources of a specific platform into a format that conforms to the interface specification. When the system runs, the corresponding adapter module is dynamically loaded according to the configuration file to achieve seamless integration of different educational resource platforms; the mechanism also includes a central configuration management system, which is used to dynamically specify which adapters need to be loaded and how to interact with different platforms to ensure the flexibility and scalability of the resource integration process. During the interaction process, the cross-platform resource association weight is determined by calculating the cosine similarity of subject categories.

[0070] The Data Collection Module uses advanced artificial intelligence algorithms, especially deep learning and natural language processing technologies, to analyze, preprocess and uniformly output multi-modal educational resources. The Data Collection Module adopts a multi-modal resource intelligent aggregation method based on knowledge points to analyze, preprocess and uniformly output multi-modal educational resources, which specifically includes the following steps:

[0071] First, the multi-modal data collection engine automatically collects text, image, audio, and video resources from different educational platforms and synchronously records the learner's like, download, and comment behavior data. Subsequently, preprocessing is carried out through data cleaning and feature extraction. The knowledge point recognition algorithm automatically recognizes the knowledge points in the text materials, and at the same time, computer vision technology is used to identify the teaching content in the video and extract the key frame timestamp information. Then, the multi-modal data fusion algorithm associates and integrates the data of different modalities according to the recognized knowledge points, and uses graph neural networks or other deep learning models to promote cross-modal spatial local feature fusion, generating a structured knowledge graph containing knowledge point difficulty labels. Finally, the intelligent aggregated processed educational resources are output in the form of structured data, which is convenient for display and use on the educational platform. At the same time, it supports exporting the processed data into multiple formats to adapt to different teaching needs and platforms, and the multiple formats include JSON, XML, and SCORM standard formats.

[0072] The innovation and feasibility of this technology are supported by the rapid development of existing artificial intelligence and machine learning technologies. Among them, natural language processing and computer vision technologies have been widely applied in multiple fields, and graph neural networks have also been verified in the academic and industrial circles.

[0073] The digital human production module includes an individual information collection unit and a model training unit connected to the individual information collection unit;

[0074] The individual information collection unit is used to collect the appearance, voice, and teaching style information of teachers, including obtaining facial feature data through a three-dimensional scanning device and collecting a voice sample library through a high-fidelity recording device; in the digital human production module of the present invention, the individual information collection unit is specifically responsible for capturing key information such as the appearance, voice, and teaching style of teachers. This unit accurately captures the facial and body features of teachers through high-precision camera equipment and three-dimensional scanning technology to construct a three-dimensional digital model. At the same time, high-quality audio collection equipment is used to record the voice samples of teachers, and voice signal processing technology is applied to extract voice features, laying a foundation for generating a realistic voice model. In addition, by analyzing the teaching videos of teachers, machine learning algorithms are used to automatically identify and extract non-verbal features reflecting the teaching style, such as body language and expressions. These collected individual information are then used to train the digital human model, enabling it to simulate the teaching experience of real teachers visually and auditorily, thus providing more abundant and personalized educational resources for learners. This process not only reflects the innovation of the present invention in technology but also demonstrates its practical application potential in the personalized and automated generation of educational resources.

[0075] The model training unit is used to train the digital human model and the voice model based on the collected individual information, and establish the mapping relationship between the teaching style and the knowledge point difficulty coefficient. The model training unit in the present invention is responsible for developing and training the digital human model and the voice model to achieve a highly personalized and natural teaching interaction experience. This unit first extracts key features from the collected individual information by using advanced computer vision and audio processing technologies. For the digital human model, the appearance features of the teacher are captured through 3D scanning and high-resolution camera technologies, and then a deep learning model is trained using this data. This model can learn and imitate the facial expressions and body movements of the teacher. In addition, combining machine learning and computer graphics technologies, the model can generate a realistic 3D digital human image.

[0076] For the voice model, the model training unit uses automatic speech recognition technology to analyze the collected teacher voice samples and extract acoustic features such as pitch, timbre, and rhythm. Then, a neural network model is trained using these features. This model can synthesize a voice output that is highly similar to the original voice of the teacher. During the model training process, third-party API services can be called, such as the text-to-speech API, to enhance the diversity and adaptability of the model, ensuring that the generated digital human voice is natural, fluent, and expressive.

[0077] In this embodiment, the model training unit also includes a feedback mechanism that allows the system to continuously optimize and adjust the digital human model and the voice model according to the user interaction data and teaching effect evaluation. Through this iterative training method, the model can gradually adapt to different teaching scenarios and the needs of learners, providing more personalized and effective teaching resources. The optimization process includes: collecting the scoring data of students on the digital human teaching and the attention concentration rate index, and updating the model parameters through the backpropagation algorithm to make the naturalness score of the digital human's voice reach more than 4.2 points (MOS standard).

[0078] This process not only reflects the innovation of the present invention in technology, but also demonstrates its practical application potential in the personalized and automated generation of educational resources, while ensuring the feasibility and efficiency of the model.

[0079] The resource generation module is a multimedia synthesis unit. The multimedia synthesis unit is used to automatically generate teaching videos and PPTs according to structured data. During the PPT generation process, the force-directed layout algorithm is adopted to control the information density not to exceed 0.35 elements per square centimeter, and knowledge point difficulty prompt marks calculated based on the learner's operation feature quantities are automatically embedded in the teaching videos. The multimedia synthesis unit in the present invention adopts advanced automation technology and can automatically generate teaching videos and PPTs according to structured teaching data. This unit uses a deep learning model to train the model to recognize teaching styles and content presentation methods by analyzing teachers' teaching videos, audio samples, and PPT templates. During the generation process, this unit can call third-party API services, such as speech synthesis and 3D model generation services, to enhance the diversity and adaptability of the content. Through intelligent algorithms, the multimedia synthesis unit can dynamically synthesize the digital image of the teacher, and generate vivid teaching videos in combination with teaching content and activities, while automatically designing and laying out PPT slides, including text, images, and charts. Finally, the generated teaching resources can be exported in various formats to adapt to different teaching platforms and devices, thus significantly improving the efficiency and quality of educational resource generation, while reducing the dependence on professional production personnel.

[0080] The dynamic adjustment module is a resource version control system. The resource version control system is used to manage different versions and changes of teaching resources. It is specifically designed to manage and track different versions and changes of teaching resources. This system ensures that the updates and iterations of each resource can be properly recorded and traced back by maintaining a detailed version record. It allows educational content developers and teachers to easily access specific versions of resources, while supporting quick rollback and update of old versions of resources. In addition, this system also provides a user-friendly interface, enabling users to intuitively compare the differences between different versions and select the resource version that best suits their needs. Through this efficient version control mechanism, the present invention can ensure the continuity and consistency of educational resources, while improving the efficiency of updating and maintaining educational resources.

[0081] The change strategies include: triggering a version rollback when the learning progress deviation of a certain knowledge point exceeds 30%, starting a content replacement process when the error rate of the knowledge point exceeds 5%, and dynamically matching five-level recommended resources according to the difference coefficient D value of the same knowledge point;

[0082] Specifically, it includes the following five-level recommendation strategies:

[0083] When the difference coefficient D < 0.2, recommend similar resources;

[0084] When 0.2 ≤ D < 0.4, recommend resources of related knowledge points;

[0085] When 0.4 ≤ D < 0.6, recommend auxiliary resources;

[0086] When 0.6 ≤ D < 0.8, supplementary resources are recommended;

[0087] When D ≥ 0.8, resources from opposing perspectives are recommended.

[0088] The user interface module includes a front - end interface and a back - end service. The front - end interface adopts responsive layout and cross - platform framework technology to ensure that the user interface can provide a consistent interaction experience on multiple devices and platforms, and integrates the WebGL engine to achieve three - dimensional rendering of the digital human teaching scenario and real - time update of the resource matching heat map. The back - end service is used to process user requests and interact with the system, adopts the gRPC framework to support high - concurrency queries of more than 1200 times per second, and controls the response latency within 50 milliseconds through Redis caching.

[0089] Specifically, the front - end interface supports multiple devices and platforms: The front - end interface design of the present invention adopts responsive layout and cross - platform framework technology to ensure that the user interface can provide a consistent interaction experience on multiple devices and platforms. Through CSS media queries and JavaScript, the adaptive adjustment of interface elements is realized. Modern front - end frameworks such as React or Vue.js are used to build reusable components, and device feature detection is carried out through the JavaScript API to apply appropriate interface styles. In addition, the progressive enhancement strategy ensures that basic functions are available on all devices, while additional functions are added for advanced devices. Code optimization and lazy - loading technology improve the loading efficiency, and extensive testing and adaptation ensure the compatibility and reliability of the interface. This design follows accessibility standards, enabling all users, including disabled persons, to access the interface conveniently, thus providing a flexible and widely - compatible front - end solution for different learning environments and user preferences.

[0090] Specifically, the back - end service is used to process user requests and interact with the system: The back - end service module design of the present invention is used to efficiently process user requests and system interactions by implementing a high - performance API server endpoint to receive and respond to HTTP requests sent by the front - end. The service adopts RESTful API or GraphQL interfaces to ensure the flexibility of front - end and back - end separation and system integration. The back - end service is responsible for executing core business logic, including data processing, transaction management, and rule application, while interacting with the database system to perform data persistence operations. In addition, the service also integrates security measures such as preventing SQL injection and cross - site scripting attacks, as well as user authentication and authorization mechanisms to ensure system security. To improve performance and user experience, the back - end service supports asynchronous processing and on - demand resource loading, reduces the system's initial loading time and resource consumption, and manages different versions and changes of teaching resources through a resource version control system to ensure the accuracy and real - time update of educational resources.

[0091] Example Two:

[0092] Refer to Figure 2 , the present invention also discloses an artificial intelligence-based automated educational resource generation method, including the following steps:

[0093] Step S1, access educational resource sites through the resource site management module, and dynamically load the adapter to parse the open API interfaces of MOOC, Coursera, and edX platforms;

[0094] Step S2, use the data collection module to collect multi-modal resources based on knowledge points, and calculate the operation feature quantity of learners and the difference coefficient D value of the same knowledge point;

[0095] Step S3, generate a digital human model and a voice model of the teacher through the digital human production module, and iteratively optimize the model parameters based on the teaching effect feedback data;

[0096] Step S4, use the resource generation module to automatically construct PPTs and generate teaching videos according to the output of the course large model, mark the difficulty level of knowledge points in the PPTs, and insert resource recommendation prompts based on the D value in the videos;

[0097] Step S5, perform real-time adjustment and intervention on the generated teaching resources through the dynamic adjustment module. When the D value is in the range of 0.6 - 0.8, supplementary resources are automatically pushed, and when the D value ≥ 0.8, resources from opposing perspectives are pushed.

[0098] Embodiment Three:

[0099] The present invention also discloses an artificial intelligence-based automated educational resource dynamic adjustment method, including the following steps:

[0100] Step A1, receive teaching feedback information (such as knowledge point error rate, click-through rate), and collect data on students' answer correct rate, resource click-through rate, and learning progress deviation in real time through the Kafka message queue;

[0101] Step A2, analyze the feedback information to determine the resources that need to be adjusted, use the random forest algorithm to classify the feedback problems, and the priority order is: knowledge point error (P1), content obsolescence (P2), interaction experience (P3);

[0102] Step A3, perform adjustment and intervention on the resources based on the analysis results, call the Git version control system to perform a differential merge operation, and update the cross-platform association weights through the resource site management module.

[0103] Embodiment Four:

[0104] The present invention also discloses an electronic device, comprising: a processor and a memory for storing executable instructions executable by the processor, wherein the processor is configured to read the executable instructions from the memory and execute the executable instructions to implement the above-mentioned automated educational resource generation method or automated educational resource dynamic adjustment method, and the electronic device is configured as a distributed computing cluster with at least 3 nodes, and the hardware specifications of a single node include: NVIDIA A100 GPU (40GB video memory), Intel Xeon Platinum 8380 processor and 512GB DDR4 memory.

[0105] Embodiment Five:

[0106] The present invention also discloses a computer-readable storage medium, on which a computer program is stored, and the computer program is configured to execute the above-mentioned automated educational resource generation method or automated educational resource dynamic adjustment method. The storage medium is pre-installed with a Docker containerized environment, the size of the image file does not exceed 850MB, and it includes the PyTorch 2.0 framework and the TensorRT acceleration library.

[0107] The implementation principle of the present invention is: The present invention discloses an automated educational resource generation and dynamic adjustment system, method, electronic device and computer-readable storage medium based on artificial intelligence. Compared with the existing online education system, it significantly improves the generation efficiency of educational resources, enhances the personalization and interactivity of resources, realizes the dynamic adjustment and optimization of resources, and provides strong technical support for online education.

[0108] The embodiments of this specific implementation manner are all preferred embodiments of the present invention, and do not limit the protection scope of the present invention accordingly. Therefore, all equivalent changes made according to the structure, shape and principle of the present invention shall be covered within the protection scope of the present invention.

Claims

1. An artificial intelligence-based automated educational resource generation and dynamic adjustment system, characterized in that, It includes a resource site management module, a data collection module, a digital human production module, a resource generation module, a dynamic adjustment module, and a user interface module; The resource site management module is used to integrate the open learning platform, third-party professional sites, and university learning resources, and optimize the cross-platform resource association relationship by analyzing the similarity of subject categories, and optimize the cross-platform resource association relationship by analyzing the similarity of subject categories; The data collection module is used to collect multi-modal resources related to knowledge points, and build a learning preference model by collecting the number of times the learner pauses the video, the number of times the progress bar is dragged, and the learning frequency data; The digital human production module is used to generate the digital human model and voice model of the teacher; The resource generation module is used to automatically construct PPTs and generate teaching videos according to the output of the course large model, and dynamically adjust the resource generation strategy based on the difficulty coefficient of the knowledge points of the learners; The dynamic adjustment module is used to perform real-time adjustment and intervention on the generated teaching resources according to teaching needs. The intervention strategies include being divided into five levels: similar resource recommendation, associated resource recommendation, auxiliary resource recommendation, supplementary resource recommendation, and opposing perspective resource recommendation according to the difference coefficient of the same knowledge point; The user interface module is used for the interaction operations between students and teachers and the system, and visually displays the heat map of the matching degree of the resource recommendation strategy in real time.

2. The automated educational resource generation and dynamic adjustment system based on artificial intelligence according to claim 1, characterized in that, The resource site management module adopts a resource integration mechanism that dynamically adapts to different platforms. The mechanism uses Adaptive technology to abstract the specific implementations of different educational resource platforms by defining a set of standardized interfaces; The mechanism includes a set of adapters. Each adapter is responsible for converting the resources of a specific platform into a format that conforms to the interface specification. When the system runs, the corresponding adapter module is dynamically loaded according to the configuration file to achieve seamless integration of different educational resource platforms; The mechanism also includes a central configuration management system, which is used to dynamically specify which adapters need to be loaded and how to interact with different platforms to ensure the flexibility and scalability of the resource integration process. The cross-platform resource association weight is determined by calculating the cosine similarity of subject categories during the interaction process.

3. An automated educational resource generation and dynamic adjustment system based on artificial intelligence according to claim 1, characterized in that, The data collection module adopts a multi-modal resource intelligent aggregation method based on knowledge points to analyze, preprocess, and uniformly output multi-modal educational resources, which specifically includes the following steps: First, use the multi-modal data collection engine to automatically collect text, images, audio, and video resources from different educational platforms, and synchronously record the learner's like, download, and comment behavior data. Subsequently, preprocess through data cleaning and feature extraction. The knowledge point recognition algorithm automatically recognizes the knowledge points in the text materials, and computer vision technology is used to identify the teaching content in the video and extract the key frame timestamp information; Next, the multi-modal data fusion algorithm associates and integrates the data of different modalities according to the identified knowledge points, and uses graph neural networks or other deep learning models to promote the cross-modal spatial local feature fusion to generate a structured knowledge graph containing knowledge point difficulty labels; Finally, the intelligent aggregated educational resources are output in the form of structured data, which is convenient for display and use on the educational platform. At the same time, it supports exporting the processed data into multiple formats to adapt to different teaching needs and platforms. The multiple formats include JSON, XML, and SCORM standard format.

4. An automated educational resource generation and dynamic adjustment system based on artificial intelligence according to claim 1, characterized in that The digital human production module includes an individual information collection unit and a model training unit connected to the individual information collection unit; The individual information collection unit is used to collect the appearance, voice, and teaching style information of teachers, including obtaining facial feature data through a three-dimensional scanning device and collecting a voice sample library through a high-fidelity recording device; The model training unit is used to train the digital human model and the voice model based on the collected individual information, and establish a mapping relationship between the teaching style and the knowledge point difficulty coefficient.

5. An automated educational resource generation and dynamic adjustment system based on artificial intelligence according to claim 4, characterized in that, The model training unit also includes a feedback mechanism that allows the system to continuously optimize and adjust the digital human model and the voice model according to user interaction data and teaching effect evaluation; The optimization process includes: collecting the scoring data of students' evaluation of digital human teaching and the attention concentration rate index, and updating the model parameters through the backpropagation algorithm to make the speech naturalness score of the digital human reach more than 4.2 points (MOS standard).

6. An artificial intelligence-based automated educational resource generation and dynamic adjustment system according to claim 1, characterized in that, The resource generation module is a multimedia synthesis unit. The multimedia synthesis unit is used to automatically generate teaching videos and PPTs according to the structured data. In the process of PPT generation, the force-directed layout algorithm is used to control the information density not to exceed 0.35 elements per square centimeter, and the knowledge point difficulty prompt marks calculated based on the learner operation feature quantity are automatically embedded in the teaching video.

7. An automated educational resource generation and dynamic adjustment system based on artificial intelligence according to claim 1, characterized in that, The dynamic adjustment module is a resource version control system. The resource version control system is used to manage different versions and changes of teaching resources. The change strategy includes: triggering version rollback when the learning progress deviation of a certain knowledge point exceeds 30%, starting the content replacement process when the error rate of the knowledge point exceeds 5%, and dynamically matching five-level recommended resources according to the difference coefficient D value of the same knowledge point.

8. An automated educational resource generation and dynamic adjustment system based on artificial intelligence according to claim 1, characterized in that, The user interface module includes a front-end interface and a back-end service. The front-end interface adopts a responsive layout and cross-platform framework technology to ensure that the user interface can provide a consistent interaction experience on multiple devices and platforms, and integrates the WebGL engine to realize the three-dimensional rendering of the digital human teaching scene and the real-time update of the resource matching heat map. The back-end service is used to process user requests and interact with the system, adopts the gRPC framework to support high-concurrency queries of more than 1200 times per second, and controls the response delay within 50 milliseconds through Redis caching.

9. An artificial intelligence-based automated educational resource generation method, characterized in that, Including the following steps: Step S1, access the educational resource site through the resource site management module, and the dynamic loading adapter parses the open API interfaces of the MOOC, Coursera, and edX platforms; Step S2, use the data collection module to collect multimodal resources based on knowledge points, and calculate the learner's operation feature quantity and the difference coefficient D value of the same knowledge point; Step S3, generate the digital human model and the voice model of the teacher through the digital human production module, and iteratively optimize the model parameters based on the teaching effect feedback data; Step S4: The resource generation module automatically constructs a PPT and generates a teaching video according to the output of the curriculum large model, marks the difficulty level of knowledge points in the PPT, and inserts resource recommendation prompts based on the D value in the video. Step S5: The dynamic adjustment module makes real-time adjustments and interventions to the generated teaching resources. When the D value is in the range of 0.6 - 0.8, supplementary resources are automatically pushed. When the D value ≥ 0.8, resources from opposing perspectives are pushed.

10. An artificial intelligence-based automated method for dynamically adjusting educational resources, characterized in that, It includes the following steps: Step A1: Receive teaching feedback information, and collect data on students' answer correct rates, resource click-through rates, and learning progress deviations in real time through the Kafka message queue. Step A2: Analyze the feedback information to determine the resources that need to be adjusted, and use the random forest algorithm to classify the feedback problems. The priority order is: knowledge point error (P1), content obsolescence (P2), interaction experience (P3). Step A3: Based on the analysis results, adjust and intervene in the resources, call the Git version control system to perform a differential merge operation, and update the cross-platform association weights through the resource site management module.

11. An electronic device, characterized in that, It includes: A processor and a memory for storing executable instructions that can be executed by the processor, where the processor is configured to read the executable instructions from the memory and execute the executable instructions to implement the method according to any one of claims 9 to 10.

12. A computer-readable storage medium, characterized in that, A computer program is stored thereon, and the computer program is configured to execute the method according to any one of claims 9 to 10.

Images