A project-based learning system

Abstract

This invention discloses a project-based learning system. The system includes a first platform, a second platform, and a server. The first platform is used for a first user to input first content, which includes at least one of voice, text, images, and video. The first content includes the first user's personalized characteristics, including at least one of voice characteristics, language habits, drawing characteristics, and video shooting characteristics. The second platform is used for a second user to pre-set second content to be displayed on the first platform. The pre-setting of the second content to be displayed on the first platform is implemented through templates pre-stored in the server, and the templates provide AI tool invocation functionality. Through the technical solution of this invention, the pre-set project templates provide AI tool invocation functionality, enabling students to generate visual and personalized works (such as AI paintings and videos) by invoking AI tools, thereby enhancing the fun and motivation of learning.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of computer technology, and in particular to a project-based learning system. Background Technology

[0002] In existing technologies, the project practice modules of project-based learning (PBL) systems are relatively rigid, mostly based on pre-set practice content and steps, requiring students to perform uniform practice tasks, lacking solutions that reflect individual student needs. Furthermore, in terms of intelligent teaching, existing recommendation systems based on learning data or knowledge graphs rely on static analysis of historical data, failing to reflect students' real-time, personalized learning status. Therefore, there is an urgent need for a project-based learning system that emphasizes practice, strong interaction, and full personalization. Summary of the Invention

[0003] This invention provides a project-based learning system that enables students to access AI tools by pre-setting project templates. This allows students to generate visual and personalized works (such as AI paintings and videos) by calling multimodal AI tools, thereby enhancing the fun and motivation of learning.

[0004] According to one aspect of the present invention, a project-based learning system is provided, comprising: a first platform, a second platform, and a server; The first platform is used to allow a first user to input first content, which includes at least one of voice, text, images, and video. The first content includes the personalized characteristics of the first user, which include at least one of voice characteristics, language habits, drawing characteristics, and video shooting characteristics. The second platform is used for a second user to pre-set the second content to be displayed on the first platform. The pre-setting of the second content to be displayed on the first platform is achieved through a template pre-stored in the server. The template provides an AI tool invocation function.

[0005] Furthermore, the second content is a video, and the AI ​​tool provided by the template is an AI question-and-answer tool. When the first user has a conversation with the AI ​​tool, the AI ​​tool provides answers based on the content of the video.

[0006] Furthermore, the AI ​​tool provides answers based on a course vector knowledge base; The AI ​​tool provides answers based on the content of the video, including: The first content is processed to obtain a first content vector; Based on the first content vector, obtain course knowledge fragments related to the first content from the course vector knowledge base; Structured prompt words are obtained based on the course knowledge fragments, the first content, and preset instructions; The structured prompts are input into the question-answering model to obtain the answer.

[0007] Furthermore, the dialogue between the first user and the AI ​​tool includes: the first user clicking on the AI ​​function icon on the first platform user interface, the AI ​​function icon being pre-set in the template, the AI ​​function icon being located in a fixed position on the first platform user interface or floating on the first platform user interface.

[0008] Furthermore, the first platform user interface includes a function button for contacting the second platform. After clicking the function button for contacting the second platform, the first platform sends first user data to the second platform, and the second platform views the first platform user interface and / or takes over the first platform.

[0009] Furthermore, the server stores the first user's personal database, which is used to generate the first user's learning profile.

[0010] Furthermore, the learning profile of the first user includes: content mastery, cognitive level, and learning interest.

[0011] Furthermore, based on the learning profile of the first user, the server uses a hybrid recommendation algorithm to recommend courses to either the first user or the second user.

[0012] Furthermore, the server employs a hybrid recommendation algorithm to recommend courses to the first user or the second user, including: The learning data of the first user is collected in real time, a feature vector set is constructed based on the learning data, and recommended courses are determined from the course vector knowledge base according to the feature vector set and recommended to the first user or the second user.

[0013] Furthermore, the server includes: The course level module is used to store and manage course content at different levels; The Project-Based Learning (PBL) project module is used to store pre-set PBL project templates; The intelligent learning guidance engine module is used for real-time guidance and personalized learning feedback; The level certification module is used to conduct online tests and generate corresponding ability level certificates based on the results after students complete the course content at different levels. The learning data analysis module is used to analyze the learning data of the first user and generate an analysis report.

[0014] This invention proposes a project-based learning system comprising a first platform 101, a second platform 102, and a server 103. Through modules such as a course grading module, a PBL project module, and an intelligent learning engine module, students can easily use AI tools to generate visual and personalized works (such as AI paintings and videos) in project-based learning, thereby enhancing the fun and motivation of learning.

[0015] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the structure of a project-based learning system according to an embodiment of the present invention; Figure 2 This is a first platform architecture diagram of a project-based learning system according to an embodiment of the present invention; Figure 3 This is a second platform architecture diagram of a project-based learning system according to an embodiment of the present invention; Figure 4 This is a server architecture diagram of a project-based learning system according to an embodiment of the present invention; Figure 5 This is a flowchart of a course recommendation process based on a hybrid recommendation algorithm in an embodiment of the present invention; Figure 6 This is a flowchart of a project-based learning method in an embodiment of the present invention. Detailed Implementation

[0018] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0019] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and their derivatives, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0020] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.

[0021] To facilitate understanding of the technical solutions disclosed in the embodiments of this disclosure, the terms involved in the embodiments of this disclosure are explained as follows.

[0022] (1) Large Model Large models refer to machine learning models with a large number of parameters and complex computational structures. They are typically composed of deep neural networks, possessing good expressive power and predictive performance, and are capable of handling complex tasks and data. Currently, large models are widely used in many fields, including natural language processing, computer vision, speech recognition, and recommender systems.

[0023] (2) Large language model Large Language Model (LLM) refers to a deep learning model trained on massive amounts of text data. This model can generate natural language text or understand the meaning of language text, and it has important applications in knowledge question answering systems.

[0024] (3) Text-to-Text Model Text-to-Text Models refer to a class of Artificial Intelligence (AI) models that take text as input and output text, with Large Language Models (LLMs) being a typical example. These models, trained on massive amounts of text data using deep learning (such as those based on the Transformer architecture), gain a deep understanding of the syntax, semantics, and context of natural language. Their core function is to generate logical, relevant, and creative text content based on input instructions or prompts, and they are widely used in tasks such as machine translation, intelligent question answering, and content completion.

[0025] (4) Image-to-text model Image-to-Text (Image-to-Text) models are a type of artificial intelligence model that takes images as input and outputs descriptive text; they are also known as visual understanding models. These models use computer vision techniques, such as Convolutional Neural Networks (CNNs) or Visual Transformers, to extract and analyze visual features from input images, identifying objects, scenes, and their relationships. Subsequently, they utilize natural language generation techniques to transform this structured visual information into fluent and accurate natural language descriptions. Their main applications include automatic image annotation and visual question answering.

[0026] (5) Text-based image model Text-to-Image Models (CTIMs) are generative artificial intelligence models that take natural language text descriptions (Prompts) as input and generate images as output. These models can understand the complex semantics of the input text and transform them into visual representations, thereby synthesizing a novel digital image that conforms to the text description. Their underlying technologies can be diffusion models or generative adversarial networks (GANs), and they have important applications in fields such as artistic creation and design assistance.

[0027] (6) Graph model Image-to-Image Models (IPMs) are a type of artificial intelligence model that takes a source image as input (often supplemented with text instructions) and generates a modified or transformed target image. Unlike text-based image models, which create images from scratch, the core task of IPMs is to edit, transform, or enhance existing images. Their applications are diverse, including converting photos to a specific artistic style (style transfer), removing or replacing elements in an image (image inpainting), and increasing image resolution (super-resolution).

[0028] (7) Graphical User Interface A graphical user interface (GUI) is a computer user interface that uses graphics to display operations.

[0029] (8) Project-based learning Project-based learning (PBL) is a learning approach that emphasizes students acquiring knowledge and skills by completing real-world projects. Students apply their learned knowledge and skills to research, plan, execute, and evaluate within these projects. This approach encourages students to apply learned concepts and skills in practice, fostering their problem-solving, innovation, and teamwork abilities.

[0030] (9) Hybrid Recommendation Algorithm Hybrid recommendation algorithms are comprehensive recommendation algorithms that combine multiple recommendation algorithms. By leveraging the advantages of each algorithm, they improve the accuracy and coverage of recommendations.

[0031] (10) Dynamic learner model A dynamic learner model is a digital model constructed based on the learning data generated by learners during the learning process. It describes and analyzes the learners' knowledge, skills, cognitive behaviors, emotional experiences, and other characteristics. As learning data accumulates, it continuously optimizes the analysis of learners' states, thereby providing learners with personalized learning paths and learning method suggestions.

[0032] (11) Enhanced search generation Retrieval-Augmented Generation (RAG) is an artificial intelligence framework that integrates retrieval techniques with generative large models. Its core mechanism lies in obtaining relevant information from external knowledge bases or real-time data sources through semantic retrieval and inputting the retrieval results as context-enhancing prompts into a large language model (LLM), thereby significantly improving the model's performance in knowledge-intensive tasks. It has important applications in fields such as open-domain question answering, multi-turn dialogue generation, long text summarization, and personalized content creation. RAG technology has many sub-types, including GraphRAG (Graph Retrieval-Augmented Generation).

[0033] Figure 1 This is a schematic diagram of the structure of a project-based learning system according to an embodiment of the present invention. This embodiment is applicable to project-based learning and the system can be implemented in software and / or hardware.

[0034] One possible implementation is, such as Figure 1 As shown, this project-based learning system includes: a first platform 101, a second platform 102, and a server 103.

[0035] The first platform 101 is used to allow a first user to input first content. The first content includes at least one of voice, text, images, and video, and incorporates the first user's personalized characteristics, including at least one of voice characteristics, language habits, drawing characteristics, and video shooting characteristics.

[0036] The second platform 102 is used by a second user to pre-set second content to be displayed on the first platform 101. The pre-setting of the second content to be displayed on the first platform 101 is achieved through a template pre-stored in the server 103, and the template provides AI tool invocation functionality.

[0037] In the specific implementation process, the first user can be a student, and the first platform 101 can be a student platform. The student platform provides functions such as course learning, project practice, level testing and work display, and interacts with students through a graphical user interface (GUI).

[0038] Figure 2 This is a first platform architecture diagram of a project-based learning system according to an embodiment of the present invention. For example... Figure 2 As shown, the present invention provides an architecture diagram of a first platform 101, comprising: Course learning module 1011 is used for the first user to learn the course content; PBL Project Practice Module 1012 is used for first users to practice PBL projects. Optionally, the first user can select a PBL project on the first platform. The assignment submission module 1013 is used for the first user to submit assignments, works, and receive feedback. The feedback can be from the second user or the project-based learning system. The intelligent learning assistant module 1014 is used to provide the first user with information on Q&A and learning suggestions. The level examination module 1015 is used for the first user to participate in a level examination, which is related to the course content being taken by the user. The learning data module 1016 is used to collect the first user's learning data, such as course completion progress, project scores, test scores, etc., and to display the first user's learning data to the first user.

[0039] The second user can be a teacher, and the second platform 102 can be a teacher platform. The teacher platform provides functions such as learning monitoring, course management and teaching intervention, supports data visualization, and is used to monitor class learning data and conduct teaching management intervention.

[0040] Figure 3 This is a second platform architecture diagram of a project-based learning system according to an embodiment of the present invention. For example... Figure 3 As shown, the present invention provides an architecture diagram of a second platform 102, comprising: The learning progress management module 1021 is used to allow a second user to view the first user's learning data, such as course completion progress, project scores, test scores, etc. Data analysis module 1022 is used to allow a second user to view the analysis results of the first user's learning data; Teaching intervention module 1023 is used to allow a second user to adjust the teaching plan according to the learning situation; The homework grading module 1024 is used for a second user to grade the homework of the first user. Optionally, the project-based learning system can use AI and other tools to automatically grade the homework of the first user. The homework grading module 1024 can be used for a second user to review and modify the results of the automated grading. Personalized guidance module 1025 is used to allow a second user to provide learning suggestions to a first user; The competition recommendation module 1026 is used by second users to recommend the works of first users to participate in competitions.

[0041] The server 103 primarily provides storage and computing functions, working in conjunction with the first platform 101 and the second platform 102 to complete various teaching and learning tasks. It stores multiple types of templates for the second user to set the content displayed on the first platform 101, such as recorded videos, live videos, and PBL (Problem-Based Learning) projects. Each template is a structured data configuration file that defines the content, steps, and interactive interface of the learning task. Optionally, it defines callable AI tool interfaces. For example, a "text creation" PBL project template integrates a button to call a large text generation model API. Teachers can quickly deploy a complete PBL project for students by selecting and configuring such a template through the second platform 102.

[0042] Figure 4 This is a server architecture diagram of a project-based learning system according to an embodiment of the present invention. Figure 4 As shown, the server 103 includes the following modules: Course level module 1031 is used to store and manage course content at different levels; PBL project module 1032 is used to store pre-set PBL project templates; The intelligent learning guidance engine module 1033 is used for real-time guidance and personalized learning feedback. The learning data analysis module 1035 is used to analyze the learning data of the first user and generate an analysis report. Large Model API Interface Module 1034 is used to provide and manage large model interfaces; The homework grading module 1036 is used to call the large model API to automatically grade homework and generate structured feedback reports; The level certification module 1037 is used to conduct online tests and generate corresponding ability level certificates based on the results after students complete the course content at different levels. The competition connection module 1038 is used to push outstanding works to the whitelist competition platform.

[0043] Specifically, a detailed introduction to each module is as follows: The course level module 1031 is used to store and manage systematic recorded courses divided into multiple levels. The course content covers the artificial intelligence knowledge system from image recognition and natural language processing to deep learning, corresponding to different school ages and cognitive levels. It ensures the systematicness, continuity and progression of the course content, and covers the basic knowledge of AI to ensure the continuity and scientific nature of the knowledge. It is suitable for students of all school ages.

[0044] Through the course grading module 1031, the present invention can realize a "dual-teacher" collaborative teaching model, in which online recorded courses provide standardized knowledge transmission, while local teachers are responsible for learning process management and personalized tutoring. PBL Project Module 1032 is used to store various types of PBL project templates and integrates an external AI large model application programming interface (API) for students to conduct project practice and generate personalized learning works such as pictures, music, videos, and articles.

[0045] Optionally, the PBL project module 1031 is used to store pre-set PBL project templates.

[0046] Optionally, each PBL project template is a structured task package that includes at least: a description of the project objectives, step-by-step implementation guidelines, configuration of one or more integrated AI tool API calls (such as required parameters and call formats), input and output specifications, and evaluation criteria.

[0047] Optionally, each PBL project is associated with a specific course stored in the course level module 1031. When the first user selects a PBL project, the interface of the first platform 101 will dynamically load the interactive elements corresponding to that PBL project, such as text boxes, upload buttons, and generate buttons. Following the instructions, the first user will use the predefined configuration in the PBL project template to call the corresponding large model API to generate personalized works, such as creating AI paintings or generating AI videos.

[0048] Optionally, the PBL project module 1032 provides functions for automatically saving, displaying, and sharing works.

[0049] The intelligent learning guidance engine module 1033 is used for real-time learning guidance, collecting student learning data, providing personalized learning suggestions and online Q&A.

[0050] Optionally, the intelligent learning engine module 1033 includes: a data acquisition unit, a vector modeling unit, and a course recommendation unit. The intelligent learning engine module 1033 integrates real-time interactive data from the first platform 101, work data from the PBL project practice module 1012, and content data from the course learning module 1011 and the tiered course module 1031 to drive automated grading and feedback. Utilizing a large language model API, it performs multi-dimensional evaluation of the PBL works submitted by the first user, such as a piece of code, an analysis report, or an AI-generated video description, considering factors such as completeness, innovation, and accuracy of technical application. It then generates structured comments and improvement suggestions, which are fed back to the first platform 101 and the second platform 102.

[0051] Optionally, the intelligent learning engine module 1033 can be used to provide the first user with real-time guidance, Q&A and personalized learning feedback through the intelligent learning assistant module 1014 of the first platform 101, to automatically grade the first user's PBL project works based on the large model API, to collect learning data (such as course completion, project score, etc.) and generate personalized reports, to store user data, to store course data, to store project data and learning records.

[0052] Optionally, deploying the intelligent learning guidance engine module 1033 for personalized learning tutoring and automated assessment may include the following aspects: (1) Construct a learning data collection system to collect multi-dimensional learning data in real time, including course completion progress, project completion quality, and test scores; (2) Provide real-time learning support through AI assistants, including Q&A on knowledge points, suggestions on learning paths and guidance on learning methods; (3) Establish an automated homework grading system, use the large model API to evaluate students' PBL works from multiple dimensions, and generate detailed feedback reports; (4) Develop a learning analysis dashboard to provide a visual representation of the overall learning situation of the first user for the second user, supporting teaching decisions and personalized intervention.

[0053] Optionally, the data acquisition unit is used to collect and structure user learning data in real time, including but not limited to: interaction data, performance data, and engagement data. The vector modeling unit is used to build and continuously update the user's personal database based on the collected learning data. The course recommendation unit is used to match the most suitable subsequent courses for the first user from the course grading module 1031 according to the dynamic learner model.

[0054] The Large Model API Interface Module 1034 is used to provide and manage large model interfaces, thereby integrating external large model capabilities and supporting multimodal content generation and automatic grading of first-user jobs.

[0055] Based on the above modules, the present invention enables students to easily call AI tools to generate visualized and personalized learning outcomes when learning high-quality and appropriately challenging content, thereby enhancing students' learning motivation and interest, and providing a good environment for practice and creation.

[0056] The learning data analysis module 1035 is used to analyze learning data, such as course completion progress, project scores, and test results, and generate analysis reports, such as analysis reports that include user learning profiles. Optionally, the learning data analysis module 1035 includes a teaching quality evaluation system, which continuously optimizes teaching content and methods by analyzing students' learning data and output.

[0057] The homework grading module 1036 is used to call the large model API to automatically grade homework and generate structured feedback reports.

[0058] The Level Certification Module 1037 is used for online level examinations and generates ability level certificates based on the examination results.

[0059] Optionally, the level certification module 1037 is used to organize online tests and generate competency level certificates based on the results after students complete a level of learning. Specifically, a level certification system is established whereby students take an online exam after completing each level of learning, and the system automatically generates competency level certificates based on the exam results.

[0060] Optionally, course levels and certification levels are linked. The course level module 1031 divides the knowledge system into tiers, with each level corresponding to specific learning objectives and ability requirements. The certification module 1037 designs a corresponding assessment system (such as theoretical exams and comprehensive reviews of practical projects). After the first user passes the assessment for that level, the system automatically generates and issues an electronic ability level certificate, realizing the visualization and standardized certification of learning outcomes.

[0061] The competition connection module 1038 is used to push outstanding works to the whitelist competition platform. Optionally, the competition connection module 1038 can construct a competition connection mechanism to connect outstanding student works with relevant whitelist competitions.

[0062] For PBL (Project-Based Learning) teaching systems, current systems are designed with independent learning, practice, testing, and management components, failing to form an effective teaching loop and personalized tutoring pathways. This embodiment provides a project-based learning system that sets up multiple functionally related modules, forming a complete incentive system of "learning-practice-testing-competition".

[0063] Among them, "learning" refers to the recorded courses on learning basic AI knowledge: multiple high-quality and interesting recorded video courses, from AI cognition enlightenment to advanced AI applications, covering multiple levels of students throughout their academic years, ensuring that students build a complete and systematic AI knowledge framework and achieve standardized knowledge input.

[0064] "Practice" refers to PBL personalized practice: multiple types of PBL practice modules, an interactive project platform integrating multimodal APIs (image generation, video recognition, etc.), with huge creative space, ensuring that each student's results are unique and realizing personalized ability transformation.

[0065] "Examination" refers to the graded examination and certification system: a scientific graded testing system and corresponding ability level certificates, combined with the assessment of basic AI knowledge and a comprehensive evaluation of AI technology application ability, to conduct a comprehensive assessment of students' abilities and achieve visualized achievement measurement.

[0066] "Competition" refers to the whitelist competitions: the advanced level content and project exercises of the course (such as Level 6 and Level 7) are designed directly to meet the participation requirements of the whitelist competitions, realizing a high-level value output.

[0067] "Management" refers to the teacher platform management system: a "dual-teacher" model of "online recorded courses + local teachers" (providing knowledge input through online courses, while local teachers are responsible for course management and Q&A to ensure teaching quality), 24 / 7 full-process guidance, and a comprehensive teaching management system integrating student management, course management, examination management, information notification, etc., to achieve large-scale quality assurance.

[0068] By integrating the "learning, practice, testing, and management" processes through an intelligent learning engine, a personalized teaching loop is formed, improving learning efficiency. The "dual-teacher" model achieves high-quality teaching through online courses and local teacher management. Online AI tutors access a large model to provide real-time guidance, as well as personalized homework correction and learning feedback.

[0069] The functional modules of the first platform 101, the second platform 102, and the server 103 correspond to each other. The first platform 101 and the second platform 102 provide data display and user interaction functions, while the server 103 provides computing resources. It will be understood by those skilled in the art that the computing power of the server 103 can be built into the first platform and the second platform respectively, thereby reducing the number of functional modules required for the server.

[0070] In the specific implementation process, optionally, the second content can be pre-recorded videos. Teachers access the course level module 1031 stored on server 103 through the teacher platform, use the course content therein to set the pre-recorded video content displayed on the student platform, which can be achieved through the course setting template stored in PBL project module 1032. Optionally, the course setting template integrates an external AI large-scale model API for students to conduct practical interactions such as knowledge Q&A. This AI large-scale model includes text-to-text, text-to-image, image-to-text, and image-to-image large-scale models, etc. Optionally, the second content can also include learning prompts, AI tool buttons, etc. When students encounter questions while learning the pre-recorded video content, they can click the AI ​​tool button, enter the question (i.e., the first content), and the AI ​​will provide an answer.

[0071] Optionally, the second content can be a PBL project. Teachers access the PBL project module 1032 stored on server 103 through the teacher platform, using the PBL project settings template and PBL project content to set the PBL project content displayed on the student platform. The second content may also include learning tips, AI tool access buttons, etc. For example, a teacher selects a "Create My AI Sci-Fi Short Film" project template, sets the project cycle and basic requirements, and then publishes it. When students are practicing PBL projects, they input at least one of voice, text, images, and videos according to the project practice guidelines—that is, the first content. The student platform, based on the template configuration, calls the corresponding AI tool APIs such as video generation and speech synthesis. The AI ​​tools generate corresponding content based on the first content input by the students. This first content is not just a simple instruction input, but also personalized data generated during the PBL project practice, reflecting the user's creativity and understanding. For example, in the "Painting Creation" PBL project, the text description entered by the user (such as "paint a wheat field under a starry sky in a certain style") includes their language habits and artistic preferences.

[0072] The technical solution of this invention provides a pre-set project template with AI tool calling function, enabling students to generate visual and personalized works (such as AI paintings and videos) by calling AI tools, thereby enhancing the fun and motivation of learning.

[0073] In one possible implementation, the second content is a video, and the AI ​​tool provided by the template is an AI question-and-answer tool. When the first user has a conversation with the AI ​​tool, the AI ​​tool provides answers based on the content of the video.

[0074] In the specific implementation process, when the first user (student) is learning video courses through the first platform 101 (student platform), the student can click the AI ​​tool button to talk to the AI ​​tool and ask questions in real time while the video is playing. The AI ​​tool will give answers based on the content of the video. The AI ​​Q&A tool provides real-time learning support, including answering questions about knowledge points, suggesting learning paths, and guiding learning methods.

[0075] One possible implementation is that the AI ​​tool provides answers based on a course vector knowledge base.

[0076] In the specific implementation process, before the AI ​​question-answering tool answers questions, it is necessary to vectorize the courseware materials, teacher lecture notes, practical guidance materials, and other content of each lesson in the course classification module 1031 to construct a proprietary course vector knowledge base. When the first user (student) asks a question to the AI ​​assistant (i.e., the AI ​​tool) through the first platform 101 (student platform), the AI ​​tool provides an answer based on the course vector knowledge base.

[0077] One possible implementation is that the AI ​​tool provides answers based on the content of the video, including: The first content is processed to obtain a first content vector; Based on the first content vector, obtain course knowledge fragments related to the first content from the course vector knowledge base; Structured prompt words are obtained based on the course knowledge fragments, the first content, and preset instructions; The structured prompts are input into the question-answering model to obtain the answer.

[0078] In practice, the process by which the AI ​​tool provides an answer based on the content of the video may specifically include the following steps: The first content can be the question asked by the first user in a dialogue with the AI ​​tool, that is, the question asked by the student to the AI ​​assistant through the student platform. The question is semantically analyzed and keywords are extracted, and the obtained keywords are vectorized to obtain the vector corresponding to the question, that is, the first content vector. Based on the first content vector, a real-time similarity retrieval is performed in the course vector knowledge base to obtain the course knowledge fragment corresponding to the question content; The course knowledge segments, the question content, and preset instructions (such as "You are an AI general education course teaching assistant") are combined into structured prompts, and the structured prompts are input into the question-answering model so that the question-answering model generates an answer based on the structured prompts.

[0079] The question-answering model can be a pre-trained large language model. Based on the structured prompts, the large language model generates an answer that incorporates course-specific content and is consistent with the teaching, and returns and displays it to the user.

[0080] Optionally, the technical solution of the present invention also includes a data accumulation and optimization step, which involves storing the user's original question, the course knowledge fragments retrieved by the system, and the final answer generated by the AI ​​as a complete question-and-answer pair in the user's personal database. This database can be used to analyze students' common questions, optimize course design, and serve as feedback data for fine-tuning and optimizing the search accuracy and prompt word templates of the AI ​​tool's search enhancement section.

[0081] Existing intelligent tutoring systems that access large-scale model APIs generate responses based on general knowledge bases, lacking the ability to perceive the current teaching progress. When students ask questions about specific course content, such as, "What is the convolutional neural network mentioned in the video?", the AI ​​assistant may fail to connect the question to the course context, providing vague or irrelevant answers, leading to misleading teaching and a fragmented learning experience.

[0082] This invention discloses a project-based learning system that employs an AI-guided learning method based on Retrieval-Augmented Generation (RAG) technology. This method enables the AI ​​assistant's answers to be strictly integrated with the specific courseware content, ensuring the accuracy of the answers and consistency with the teaching. It can improve the accuracy and contextual relevance of large language model dialogues in educational scenarios.

[0083] Through the RAG-based course context-aware AI-guided learning method in this embodiment of the invention, the AI ​​assistant's answers are strictly limited and dependent on the courseware content, effectively avoiding the "illusion" phenomenon, ensuring the accuracy and authority of the teaching content, making the AI ​​assistant an organic extension of the course itself rather than an external tool, and providing a seamless and coherent learning experience.

[0084] In the above process, all interactive data is stored in a structured manner, which not only optimizes the learning path of individual students, but also provides a data foundation for continuous optimization of the teaching content and AI performance of the entire system.

[0085] One possible implementation is that the AI ​​tool provides an answer based on the content of the video, including: providing an answer based on the content of the video within a predetermined duration; the starting point of the predetermined duration is determined based on the time of the question.

[0086] This can be understood as follows: when a user asks a question to the AI ​​assistant through the student platform, the question time is taken as the starting point of the predetermined duration. The AI ​​assistant gives an answer based on the content of the video within the predetermined duration. Optionally, the content of the video within the predetermined duration is obtained forward and backward based on the question time, and an answer is given accordingly. Optionally, the predetermined duration is 1 minute.

[0087] One possible implementation is that the first user interacts with the AI ​​tool by clicking an AI function icon on the user interface of the first platform 101. The AI ​​function icon is pre-set in the template and is located in a fixed position on the user interface of the first platform 101 or is floating on the user interface of the first platform 101.

[0088] Optionally, the dialogue between the first user and the AI ​​tool may include: the first user inputting voice commands. Alternatively, the dialogue between the first user and the AI ​​tool may also include: the first user inputting text, images, videos, or other commands; this embodiment does not limit this.

[0089] By setting up AI function buttons on the user interface, students can easily ask questions. Based on the course vector knowledge base, the AI ​​tool can better understand students' ambiguous questions in conjunction with the course content, thereby helping students resolve their doubts in a timely and effective manner.

[0090] One possible implementation is that the user interface of the first platform 101 includes a function button for contacting the second platform 102. After clicking the function button, the first platform 101 sends first user data to the second platform 102. The first user data includes the first user's learning data on the first platform 101 (such as course completion progress, project scores, test scores, etc.), the first user's registration information on the first platform 101, the first user's personal information data pre-entered by the second user, the device information of the first platform 101, and the content of the user interface of the first platform 101.

[0091] Optionally, after clicking the function button to contact the second platform 102, the second platform 102 can view the user interface of the first platform 101 and / or take over the first platform 101. The second platform 102 can obtain the page information of the first platform 101 and remotely operate the first platform 101.

[0092] In one possible implementation, the server 103 stores a personal database of the first user, which is used to generate a learning profile of the first user.

[0093] Specifically, the course content will be accurately pushed based on the first user's personal database.

[0094] Optionally, the learning profile of the first user includes: content mastery, cognitive level, and learning interest.

[0095] Optionally, the server 103 recommends courses to the first user or the second user based on the learning profile of the first user and using a hybrid recommendation algorithm.

[0096] Existing teaching systems often rely on single test scores or completion progress for course recommendations. Their simple recommendation logic fails to accurately reflect students' true knowledge mastery, learning interests, and cognitive development potential. This is particularly problematic in large classes, where it hinders personalized learning, making it difficult to automatically provide suitable advanced challenges for high-achieving students or to offer targeted reinforcement for students with weaker foundations.

[0097] To address the aforementioned issues, this invention provides a personalized course recommendation scheme based on multi-dimensional learning data analysis. Its core lies in using a dynamic learner model to comprehensively depict students' learning behavior and employing a hybrid recommendation algorithm to achieve precise delivery of course content.

[0098] Optionally, the server 103 employs a hybrid recommendation algorithm to recommend courses to the first user or the second user, including: The learning data of the first user is collected in real time, a feature vector set is constructed based on the learning data, and recommended courses are determined from the course vector knowledge base according to the feature vector set and recommended to the first user or the second user.

[0099] Figure 5 This is a flowchart illustrating a course recommendation process based on a hybrid recommendation algorithm, as described in this embodiment of the invention. The course recommendation process based on the hybrid recommendation algorithm in this embodiment is as follows: Figure 5 As shown: S502: Collect and structure the user's learning data.

[0100] The learning data includes, but is not limited to: a) Interaction data, such as the content of questions asked in the user interface of the first platform 101, the frequency of questions asked, and the number of dialogue rounds with the AI ​​assistant; b) Performance data, such as the accuracy rate of in-course quizzes, answering time, and automated grading scores of PBL project works; c) Engagement data, such as the time spent in each course chapter and the completion rate of video courses.

[0101] S504: Based on the collected learning data, construct and continuously update the user's personal database.

[0102] This module quantifies the user state by calculating at least one of the following learner feature vectors: a) Knowledge Mastery Vector: Based on test accuracy and answer patterns, it assesses the degree of mastery of specific knowledge points; b) Cognitive Level Vector: Based on the complexity of the questions and the innovation of the PBL projects, natural language processing technology is used to assess the students' thinking level. c) Learning Interest Vector: Based on students' engagement in different types of courses and projects, their interest preferences are identified. This can be understood as the feature vector set including a knowledge mastery vector, a cognitive level vector, and a learning interest vector.

[0103] S506: Use a hybrid recommendation algorithm to match the most suitable subsequent courses for the user from the course grading module.

[0104] The hybrid recommendation algorithm is a combination of collaborative filtering and knowledge state assessment. Its execution logic includes: First, based on knowledge state assessment, mandatory courses are recommended to consolidate weak knowledge points or connect to the next knowledge unit; second, based on collaborative filtering, learners with similar feature vectors are found in the user group, and their completed and highly rated courses are recommended as extension courses; finally, a potential activation mechanism is introduced. When the system identifies that a student's cognitive level vector is consistently higher than the current course level threshold, and the learning interest vector shows a strong willingness to explore, higher-level course recommendations are automatically triggered, pushing higher-level core courses or challenging PBL projects to the student platform.

[0105] The technical solution of this invention constructs a dynamic learner model through multi-dimensional data and employs a hybrid recommendation algorithm to achieve a shift from "one-size-fits-all" to "personalized" course recommendations. Its technical advantages are: a) Accuracy: transcending a single performance indicator, it achieves more precise course matching from three dimensions: knowledge, cognition, and interest; b) Adaptability: the model dynamically updates with the student's learning progress, and the recommendation results are adjusted in real time, forming an adaptive closed loop of "learning-assessment-recommendation-relearning"; c) Incentive: through a potential-stimulating mechanism, it automatically identifies and guides outstanding students into advanced learning stages, effectively solving the problem of "under-challenged" students in large-class teaching and maximizing the value of personalized education.

[0106] As an exemplary description of an embodiment of the present invention Figure 6 This is a flowchart of a project-based learning method according to an embodiment of the present invention. For example... Figure 6 As shown, the project-based learning process based on this system can be described as follows: Step a, Course Learning: The first user accesses the course level module 1031 through the first platform and selects the corresponding level of recorded course for learning.

[0107] Step b, Project Practice: After each lesson, the system pushes a corresponding project through the PBL project module 1032; the first user selects the project type through the GUI and calls the multimodal AI interface set to generate personalized works (for example, inputting a text description and outputting an AI painting through the image generation API).

[0108] Step c, Intelligent Guidance and Grading: The intelligent guidance engine module 1033 monitors learning progress in real time. Specifically, guidance: the intelligent guidance assistant module 1014 provides learning suggestions based on learning data. Grading: After submission, the engine calls the large model API for automated grading and generates a feedback report.

[0109] Step d, Level Testing and Certification: After completing a level of learning, students take an online level exam; the system generates a competency level certificate based on the exam results.

[0110] Step e, Teaching Management: Teachers can view class learning data (such as course completion rate and project score) through the teacher platform and intervene or adjust the teaching plan accordingly.

[0111] The entire process forms a closed loop of "learning, practicing, testing, and managing," and expands the "competition" component through level certificates and whitelisted competition links.

[0112] The technical solution of this invention, by constructing a systematic curriculum grading system, integrating a multimodal AI interface into a PBL project practice environment, deploying an intelligent learning engine, and establishing a complete teaching management and certification mechanism, achieves systematic, interactive, and personalized AI general education for all school ages, effectively solving the technical problems of fragmented curriculum content, disconnected teaching links, and low visualization of learning outcomes in existing technologies.

[0113] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0114] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A project-based learning system, characterized in that, The system includes: a first platform, a second platform, and a server; The first platform is used to allow a first user to input first content, which includes at least one of voice, text, images, and video. The first content includes the personalized characteristics of the first user, which include at least one of voice characteristics, language habits, drawing characteristics, and video shooting characteristics. The second platform is used for a second user to pre-set the second content to be displayed on the first platform. The pre-setting of the second content to be displayed on the first platform is achieved through a template pre-stored in the server. The template provides an AI tool invocation function.

2. The system according to claim 1, characterized in that, The second content is a video, and the AI ​​tool provided by the template is an AI question-and-answer tool. When the first user has a conversation with the AI ​​tool, the AI ​​tool gives an answer based on the content of the video.

3. The system according to claim 2, characterized in that, The AI ​​tool provides answers based on a course vector knowledge base; The AI ​​tool provides answers based on the content of the video, including: The first content is processed to obtain a first content vector; Based on the first content vector, obtain course knowledge fragments related to the first content from the course vector knowledge base; Structured prompt words are obtained based on the course knowledge fragments, the first content, and preset instructions; The structured prompts are input into the question-answering model to obtain the answer.

4. The system according to claim 2, characterized in that, The first user interacts with the AI ​​tool by clicking an AI function icon on the first platform user interface. The AI ​​function icon is pre-set in the template and is located in a fixed position on the first platform user interface or floats on the first platform user interface.

5. The system according to any one of claims 1-4, characterized in that, The first platform user interface includes a function button to contact the second platform. After clicking the function button to contact the second platform, the first platform sends first user data to the second platform, and the second platform views the first platform user interface and / or takes over the first platform.

6. The system according to claim 1, characterized in that, The server stores the first user's personal database, which is used to generate the first user's learning profile.

7. The system according to claim 1, characterized in that, The learning profile of the first user includes: content mastery, cognitive level, and learning interest.

8. The system according to any one of claims 6 or 7, characterized in that, Based on the learning profile of the first user, the server uses a hybrid recommendation algorithm to recommend courses to either the first user or the second user.

9. The system according to claim 8, characterized in that, The server employs a hybrid recommendation algorithm to recommend courses to either the first user or the second user, including: The learning data of the first user is collected in real time, a feature vector set is constructed based on the learning data, and recommended courses are determined from the course vector knowledge base according to the feature vector set and recommended to the first user or the second user.

10. The system according to claim 1, characterized in that, The server includes: The course level module is used to store and manage course content at different levels; The Project-Based Learning (PBL) project module is used to store pre-set PBL project templates; The intelligent learning guidance engine module is used for real-time guidance and personalized learning feedback; The level certification module is used to conduct online tests and generate corresponding ability level certificates based on the results after students complete the course content at different levels. The learning data analysis module is used to analyze the learning data of the first user and generate an analysis report.

Images