A knowledge graph-based intelligent education system and method

Abstract

This invention relates to the field of smart education technology and discloses a knowledge graph-based smart education system and method. The knowledge graph-based smart education system includes: a multi-source data acquisition module, a knowledge graph construction and updating module, a personalized teaching push module, a learning dynamic analysis module, an intelligent interaction module, a permission management module, and a data storage module. These modules work collaboratively to form a complete smart education ecosystem. This invention addresses the problem of fragmented knowledge by constructing a systematic knowledge system. Through the knowledge graph construction and updating module, scattered teaching knowledge points are structurally modeled, clearly presenting the pre-dependencies, parallel relationships, and extensions between knowledge points. This breaks down the knowledge barriers of traditional teaching, helps students grasp the overall logic of the knowledge system, improves knowledge absorption efficiency and application ability, and solves the problem of weak knowledge connections in existing education systems.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent education, and specifically relates to an intelligent education system and method based on a knowledge graph. Background Art

[0002] With the deep promotion of the digital transformation of education, intelligent education has become an important direction to promote the high-quality development of education. Currently, existing intelligent education systems mostly have the following technical defects: First, the knowledge presentation is fragmented, and no systematic knowledge association is formed, resulting in students' difficulty in grasping the overall logic of the knowledge system and low learning efficiency; Second, the teaching personalization is insufficient, and the "one-size-fits-all" teaching mode is mostly adopted, and it is impossible to formulate targeted teaching plans according to students' knowledge levels and learning preferences, making it difficult to meet the learning needs of different students; Third, the analysis of students' learning situations is not accurate enough. Only simple answering data can be used for analysis, and it is impossible to identify students' weak knowledge chains in combination with knowledge association relationships, and the intervention measures lack pertinence; Fourth, the teaching ecological closed-loop is missing, and there is no effective linkage among teaching data, learning data, and knowledge data, making it impossible to realize the full-process optimization of "teaching, learning, evaluation, and improvement".

[0003] As a structured knowledge representation method, a knowledge graph can clearly present the association relationships between knowledge points, providing technical support for the personalization and precision of intelligent education. Existing education-related technologies based on knowledge graphs mostly focus on the construction of knowledge graphs or the intelligence of a single link, such as only realizing the associated display of knowledge points or simple exercise pushing, without forming a complete closed-loop from knowledge graph construction, personalized teaching, learning situation analysis to graph update and ecological optimization, and there are deficiencies in aspects such as the dynamic update of knowledge graphs, the identification of weak knowledge chains, and the precise generation of personalized teaching plans, making it impossible to give full play to the core role of knowledge graphs in intelligent education and difficult to meet the actual needs of the digital transformation of education.

[0004] Therefore, it is of great practical significance and application value to develop an intelligent education system and method based on a knowledge graph that can achieve systematic knowledge modeling, personalized teaching adaptation, dynamic learning situation monitoring, and teaching ecological closed-loop.

[0005] Based on this, an intelligent education system and method based on a knowledge graph are designed. Summary of the Invention

[0006] In view of the above situation, to overcome the defects of the prior art, the present invention provides an intelligent education system and method based on a knowledge graph, effectively solving the problems raised in the above background.

[0007] To achieve the above object, the present invention provides the following technical solution: An intelligent education system based on a knowledge graph, comprising: The multi-source data acquisition module is used to collect various types of data during the teaching process, including teacher teaching data, student learning data, teaching resource data, and environmental data. The teacher teaching data includes lesson plans, courseware, teaching videos, and teaching evaluation records. The student learning data includes learning time, answer data, wrong answer records, learning preferences, and classroom interaction data. The teaching resource data includes textbook knowledge points, test questions, micro-lessons, and supplementary materials. The environmental data includes classroom atmosphere data and equipment operation data. The knowledge graph construction and update module is used to construct an initial knowledge graph based on teaching resource data and student learning data collected by the multi-source data acquisition module, and dynamically update the knowledge graph according to real-time data. The initial knowledge graph includes knowledge point nodes, knowledge point association edges, knowledge point attributes, and student knowledge mastery status nodes. The knowledge point association edges include prerequisite dependencies, parallel relationships, and extension relationships. The knowledge point attributes include difficulty coefficient, importance, and appropriate learning stage. The student knowledge mastery status nodes are associated with corresponding knowledge point nodes and labeled with mastery level values. The personalized teaching recommendation module is used to analyze students' knowledge weaknesses and learning preferences based on the knowledge graph generated by the knowledge graph construction and update module, combined with student learning data, to generate personalized teaching plans and learning resource recommendation lists. The personalized teaching plans include the order of knowledge point learning, the allocation of learning time, and the exercise training plan. The learning resource recommendation list matches corresponding micro-lessons, exercises, and extension materials according to students' weak knowledge points and learning preferences. The learning dynamic analysis module is used to collect student learning data in real time, combine it with knowledge graphs, analyze students' knowledge mastery progress, the evolution of weak knowledge points, and learning behavior patterns, generate learning analysis reports, identify student learning abnormalities and trigger intervention prompts. The learning analysis reports include individual student learning reports and class-wide learning reports. The learning abnormalities include lagging learning progress, persistently low mastery of knowledge points, and abnormal learning behaviors. The intelligent interaction module is used to realize two-way interaction between teachers, students and the system. It includes functions such as editing teaching plans, viewing student learning status and adjusting teaching on the teacher's side, viewing learning resources, answering exercises and submitting questions on the student's side, as well as online interaction and Q&A between teachers and students. The permission management module is used to manage the permissions of system users in a hierarchical manner. The users include administrators, teachers, and students. Administrators have full system operation permissions, teachers have the permissions to view teaching data, edit teaching plans, and intervene in learning progress for their corresponding classes, and students only have the permissions to view personal learning data, access learning resources, and submit questions. The data storage module stores all data collected by the multi-source data acquisition module, knowledge graph data generated by the knowledge graph construction and update module, teaching plan data generated by the personalized teaching push module, learning progress report data generated by the learning progress dynamic analysis module, and all log data during system operation. It adopts a distributed storage architecture to ensure data security and efficient access.

[0008] Preferably, the knowledge graph construction and updating module includes a knowledge extraction unit, a knowledge fusion unit, a knowledge storage unit, and a graph updating unit; The knowledge extraction unit is used to extract knowledge point entities, knowledge point attributes, and knowledge point relationships from teaching resource data. It employs deep learning-based entity extraction and relationship extraction algorithms to ensure an extraction accuracy of no less than 98%. The knowledge fusion unit is used to deduplicatize and standardize the extracted knowledge point entities, attributes and relationships, solve the problem of knowledge redundancy and conflict in multi-source data, and map knowledge of different formats and sources to the same knowledge system. The knowledge storage unit is used to store knowledge graph data using a graph database, supporting quick querying and modification of knowledge point nodes and related edges, and storing historical version data of the knowledge graph for easy backtracking. The knowledge graph update unit is used to periodically update the knowledge graph based on real-time collected student learning data, teacher teaching data, and newly added teaching resource data. The update cycle can be set to 1-7 days, and the update content includes adjusting the relationship between knowledge points, updating the attributes of knowledge points, and updating the student knowledge mastery status nodes.

[0009] Preferably, the personalized teaching push module includes a student profile building unit, a weakness identification unit, a teaching plan generation unit, and a resource matching unit; The student profile building unit is used to build a student profile based on student learning data, including learning preferences, knowledge mastery level, learning ability, and learning habits, and to label the feature values ​​of each dimension using a quantitative scoring method. The weak point identification unit is used to combine knowledge graph and student learning data, and identify the student's weak knowledge points by comparing the standard value of the knowledge point mastery with the student's actual mastery value. At the same time, it analyzes the pre-dependent knowledge points of the weak knowledge points to form a weak knowledge chain. The teaching plan generation unit is used to generate personalized learning paths and teaching plans based on student profiles and weak knowledge chains, specifying the learning time, learning methods and assessment standards for each knowledge point; The resource matching unit is used to match corresponding learning resources from the teaching resource library according to the personalized teaching plan and students' learning preferences, so as to ensure that the difficulty of the resources is suitable for the students' knowledge level and the type of resources is matched with the students' learning preferences.

[0010] Preferably, the learning dynamic analysis module includes a real-time data acquisition unit, a learning calculation unit, an anomaly identification unit, and a report generation unit; The real-time data acquisition unit is used to collect students' learning behavior data and answer data in real time, with a collection frequency of no less than once per minute to ensure the real-time nature of the data. The learning progress calculation unit is used to calculate the mastery level value of each knowledge point of the student based on the knowledge graph. The mastery level value is quantified from 0 to 100 points. At the same time, it calculates the student's learning progress, answer accuracy rate, and error repetition rate, among other learning progress indicators. The anomaly identification unit is used to set learning indicator thresholds. When a student's learning indicator exceeds the threshold range, it is determined to be a learning anomaly, triggering an intervention prompt. The threshold can be adjusted by the teacher according to the actual situation of the class. The report generation unit is used to periodically generate learning analysis reports. A single student's learning report includes an overview of their knowledge mastery, weak knowledge points, and learning suggestions. A class-wide learning report includes the class's average mastery level, common weak knowledge points, and teaching optimization suggestions.

[0011] A knowledge graph-based smart education method includes the following steps: Step 1: Knowledge Graph Construction. The knowledge extraction unit of the knowledge graph construction and update module extracts knowledge point entities, attributes, and relationships from teaching resource data collected by the multi-source data acquisition module. The knowledge fusion unit deduplicates and standardizes the extracted knowledge to resolve redundancy and conflicts. The knowledge storage unit uses a graph database to store the processed knowledge, constructing an initial knowledge graph. This initial knowledge graph includes knowledge point nodes, edges, attributes, and initial student knowledge mastery status nodes. Step 2: Multi-source data acquisition and preprocessing. Through the multi-source data acquisition module, teacher teaching data, student learning data, teaching resource data, and environmental data are collected. The collected data is cleaned, deduplicated, and standardized to remove invalid data. Data of different formats is converted into a unified format and stored in the data storage module to provide data support for subsequent teaching push and learning analysis. Step 3: Personalized teaching strategy generation. The student profile building unit in the personalized teaching push module constructs student profiles based on preprocessed student learning data; the weakness identification unit identifies students' weak knowledge points and weak knowledge chains by combining the initial knowledge graph; the teaching plan generation unit generates personalized teaching plans based on student profiles and weak knowledge chains; and the resource matching unit matches corresponding learning resources and pushes them to the student's end, while simultaneously synchronizing the personalized teaching plans to the teacher's end. Step Four: Dynamic Analysis and Intervention of Learning Progress. The dynamic learning progress analysis module uses a real-time data collection unit to gather students' behavioral and answer data during the learning process. The learning progress calculation unit, based on the knowledge graph and the real-time collected data, calculates learning progress indicators such as students' mastery of knowledge points and learning progress. The anomaly identification unit compares learning progress indicator thresholds to identify learning anomalies and triggers intervention prompts, which are then pushed to both teachers and students. The report generation unit regularly generates learning progress analysis reports for individual students and the entire class, providing a basis for teachers to adjust their teaching and for students to improve their learning. Teachers can view the learning progress reports through the intelligent interaction module, adjust teaching plans and personalized push strategies based on students' learning anomalies and weaknesses, and students can engage in targeted learning based on the learning progress reports and pushed resources. Step 5: Iterative update of the knowledge graph. Through the graph update unit of the knowledge graph construction and update module, real-time teaching data and student learning data are collected regularly. The changes in the relationship between knowledge points, the adjustment of knowledge point attributes, and the updates of students' knowledge mastery status are analyzed. The knowledge graph is iteratively optimized, and knowledge point nodes, related edges, attributes, and student knowledge mastery status nodes are updated to ensure that the knowledge graph is synchronized with the actual teaching. At the same time, the historical versions of the knowledge graph are stored for easy backtracking and querying. Step Six: Optimizing the Teaching Ecosystem. Based on the learning analysis report and updated knowledge graph data, teachers optimize teaching plans and resources through the intelligent interaction module, students adjust their learning plans according to personalized push notifications and learning feedback, and administrators maintain system operation and monitor data security through the permission management module. The system accumulates teaching and learning data through the data storage module, forming a closed-loop teaching ecosystem of "data collection - knowledge graph construction - teaching push - learning analysis - knowledge graph update - ecosystem optimization", continuously improving teaching efficiency and learning outcomes.

[0012] Preferably, in step one, the deep learning algorithms used by the knowledge extraction unit include the BERT entity extraction model and the CNN relation extraction model, wherein the extraction accuracy of the BERT entity extraction model is not less than 98%, and the extraction accuracy of the CNN relation extraction model is not less than 97%; the knowledge fusion unit uses a deduplication algorithm based on semantic similarity, with the semantic similarity threshold set to 0.95. When the semantic similarity of two knowledge point entities is ≥0.95, they are determined to be the same entity and are merged.

[0013] Preferably, in step two, data preprocessing includes the following sub-steps: S2.1: Data cleaning, removing missing values, outliers and duplicate data. Data with more than 5% missing values ​​are directly removed, while data with ≤5% missing values ​​are filled with mean or interpolation. S2.2: Data standardization, converting numerical data of different magnitudes into standardized data in the 0-1 range, and converting text data into a recognizable encoding format to ensure data consistency; S2.3: Data is classified and stored separately. The pre-processed teacher teaching data, student learning data, teaching resource data and environmental data are stored in the corresponding partitions of the data storage module for easy retrieval and retrieval later.

[0014] Preferably, in step four, the learning indicator thresholds include the learning progress threshold, the knowledge point mastery threshold, and the answer accuracy threshold. The learning progress threshold is set to ±10% of the teaching progress of the corresponding grade level, the knowledge point mastery threshold is set to 60 points, and the answer accuracy threshold is set to 50%. When a student's learning progress is lower than 90% of the teaching progress, the knowledge point mastery is lower than 60 points, or the answer accuracy is lower than 50%, it is judged as an abnormal learning situation, and an intervention prompt is triggered.

[0015] Preferably, in step five, the update cycle of the knowledge graph can be set by the administrator or the teacher, with a default update cycle of 3 days. The update content includes: adding knowledge point nodes and related edges corresponding to the new teaching resources, adjusting the difficulty coefficient and importance of the knowledge points, updating the student knowledge mastery status nodes, and optimizing the knowledge point relationships. Among them, the student knowledge mastery status nodes are updated in real time based on the student's latest answer data and learning time.

[0016] Preferably, in step six, the optimization of the teaching ecosystem also includes system performance optimization. This involves improving data access speed through the distributed storage architecture of the data storage module, ensuring that the system's concurrent access volume is no less than 1,000 users online simultaneously; regularly updating user permissions through the permission management module to ensure system data security; and optimizing the interactive interface through the intelligent interaction module to improve the user experience.

[0017] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention solves the problem of fragmented knowledge and constructs a systematic knowledge system. Through the knowledge graph construction and updating module, it structures and models scattered teaching knowledge points, clearly presents the pre-dependencies, parallels, extensions and other relationships between knowledge points, breaks down the knowledge barriers of traditional teaching, helps students grasp the overall logic of the knowledge system, improves the efficiency of knowledge absorption and knowledge application, and solves the problem of weak knowledge connections in the existing education system. 2. This invention enables personalized teaching adaptation to meet differentiated learning needs. Based on knowledge graphs and student learning data, it constructs accurate student profiles, identifies students' weak knowledge points and weak knowledge chains, and generates targeted personalized teaching plans and learning resource recommendation lists. It abandons the "one-size-fits-all" teaching model, adapts to the different knowledge levels, learning preferences and learning abilities of different students, makes teaching more targeted, effectively improves learning outcomes, and solves the defects of insufficient personalized teaching. 3. This invention improves the accuracy of learning analysis and enables precise intervention. By combining knowledge graphs and real-time collected student learning data, it can not only analyze the mastery of individual knowledge points, but also trace the prerequisites of weak knowledge points, identify weak knowledge chains, monitor learning status in real time, identify learning anomalies and trigger intervention prompts, and generate a refined learning analysis report. This provides a precise basis for teachers to adjust their teaching and students to improve their learning, and solves the problems of inaccurate learning analysis and lack of targeted intervention. 4. This invention constructs a closed-loop teaching ecosystem, realizing full-process intelligence and forming a complete closed loop of "data collection - graph construction - teaching push - learning analysis - graph update - ecosystem optimization". It realizes the effective linkage of teaching data, learning data, and knowledge data, connects the entire process of "teaching, learning, evaluation, and correction", and enables each module to work collaboratively to continuously optimize the teaching ecosystem, promote the continuous improvement of teaching efficiency and learning outcomes, and solve the technical problem of the lack of a closed-loop teaching ecosystem. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0019] In the attached diagram: Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0020] 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. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0021] Depend on Figure 1 The present invention relates to a knowledge graph-based smart education system, comprising: a multi-source data acquisition module, a knowledge graph construction and updating module, a personalized teaching push module, a learning dynamic analysis module, an intelligent interaction module, a permission management module, and a data storage module. The modules work together to form a complete smart education ecosystem.

[0022] 1. Multi-source data acquisition module It forms the system's data foundation, used to collect various types of data during the teaching process, ensuring the comprehensiveness and real-time nature of the data, and providing support for subsequent knowledge graph construction, personalized teaching delivery, and learning analysis. Specific data collected includes: 1) Teacher teaching data: including lesson plans, courseware (PPT, documents, etc.), teaching videos, classroom teaching recordings, teaching evaluation records (evaluation of students' classroom performance and homework completion), teaching plans, etc. 2) Student learning data: including students' learning time (learning time for each knowledge point), answer data (answers to exercises, answering time, and records of wrong answers), learning preferences (preferred types of learning resources, learning time, and learning methods), classroom interaction data (raising hands to speak, asking questions, and participating in discussions), and homework completion status, etc. 3) Teaching resource data: including textbook knowledge points, test questions (multiple choice, fill-in-the-blank, short answer, etc.), micro-lecture videos, supplementary materials (literature, case studies, experimental videos), teaching courseware, etc., covering teaching content for different grade levels and different subjects; 4) Environmental data: including classroom atmosphere data (student attention concentration, classroom interaction frequency), equipment operation data (system server operation status, terminal device connection status), etc.

[0023] This module adopts a multi-interface data acquisition method, supporting the connection with teaching terminals, learning terminals, courseware management systems, answer systems and other devices and systems. The acquisition frequency is no less than once per minute to ensure the real-time nature of the data. At the same time, the acquired data is initially screened to remove obviously invalid data (such as garbled characters, duplicate submitted empty data).

[0024] 2. Knowledge Graph Construction and Update Module It is the core module of the system, responsible for the initial construction and dynamic iterative updates of the knowledge graph, realizing the systematic modeling and real-time optimization of knowledge, and specifically includes four sub-units: 1) Knowledge Extraction Unit: Employing the BERT entity extraction model and CNN relation extraction model based on deep learning, this unit extracts knowledge point entities (such as "quadratic equation" and "Newton's first law"), knowledge point attributes (such as difficulty level, importance, appropriate grade level, and knowledge point type), and knowledge point relationships (such as the parallel relationship between "quadratic equation" and "linear equation", the pre-dependency relationship between "Newton's first law" and "inertia", and the extended relationship between "photosynthesis" and "ecosystem") from teaching resource data collected by the multi-source data acquisition module. The extraction accuracy of the BERT entity extraction model is no less than 98%, and the extraction accuracy of the CNN relation extraction model is no less than 97%, ensuring the accuracy of knowledge extraction.

[0025] 2) Knowledge Fusion Unit: This unit is used to deduplicatize and standardize the extracted knowledge point entities, attributes, and relationships to solve the problems of knowledge redundancy and conflict in multi-source data. It adopts a semantic similarity-based deduplication algorithm with a semantic similarity threshold of 0.95. When the semantic similarity of two knowledge point entities is ≥0.95, they are determined to be the same entity and merged. It standardizes and converts knowledge point attributes in different formats (such as difficulty coefficients "easy / medium / difficult" and "1 / 2 / 3") into quantitative values ​​or unified descriptions. It verifies conflicting knowledge point relationships (such as different descriptions of the same knowledge point relationship in different textbooks) in conjunction with authoritative teaching resources to determine unique relationships.

[0026] 3) Knowledge storage unit: The knowledge graph data is stored using a graph database (such as Neo4j). Graph databases have efficient node query and relationship traversal capabilities, which can quickly respond to the needs of knowledge point association query and weak knowledge point tracing. At the same time, the historical version data of the knowledge graph is stored, with each version corresponding to an update cycle, which makes it easy to backtrack and query the status of the knowledge graph at different times and provide historical data support for teaching optimization.

[0027] 4) Knowledge Graph Update Unit: Based on real-time collected student learning data, teacher teaching data, and newly added teaching resource data, the knowledge graph is periodically iterated and updated. The update cycle can be set by the administrator or teacher, with a default update cycle of 3 days. The update content specifically includes: adding knowledge point nodes and related edges corresponding to new teaching resources (such as adding knowledge points corresponding to new micro-lessons and their relationships with other knowledge points); updating knowledge point attributes (such as adjusting the difficulty coefficient of knowledge points based on students' correct answer rates and adjusting the importance of knowledge points based on the teaching syllabus); updating student knowledge mastery status nodes (updating the student's mastery level value for each knowledge point based on the latest student answer data and learning time); and optimizing the relationships between knowledge points (such as supplementing or adjusting the strength of relationships between knowledge points based on student learning feedback).

[0028] 3. Personalized teaching push module Based on knowledge graphs and student learning data, personalized teaching plans are generated and learning resources are precisely delivered to meet the learning needs of different students. This includes four sub-units: 1) Student Profile Construction Unit: Based on student learning data preprocessed by the multi-source data acquisition module, a student profile is constructed that includes learning preferences, knowledge mastery level, learning ability, and learning habits. Quantitative scoring methods are used to label the feature values ​​of each dimension, such as learning preference dimension (micro-lesson preference value, exercise preference value, document preference value), knowledge mastery level dimension (mastery level of each subject and knowledge point), learning ability dimension (answering speed, knowledge point comprehension ability, self-learning ability), and learning habits dimension (study duration, study frequency, and error correction habits), forming a comprehensive and accurate student profile.

[0029] 2) Weakness Identification Unit: Combining knowledge graphs and student learning data, this unit identifies students' weak knowledge points by comparing the standard value (60 points) of knowledge point mastery with the students' actual mastery value. At the same time, through the correlation of the knowledge graph, it traces the prerequisite knowledge points of the weak knowledge points to form a weak knowledge chain (e.g., if a student has a weak grasp of "solving quadratic equations in one variable", it traces the prerequisite knowledge points "solving linear equations in one variable" and "factorization" to form a weak knowledge chain), providing a basis for the generation of personalized teaching plans.

[0030] 3) Teaching Plan Generation Unit: Based on student profiles and weak knowledge chains, generate personalized learning paths and teaching plans; specifically including: knowledge point learning order (prioritize learning prerequisite knowledge points in the weak knowledge chain, and then learn the target weak knowledge points), learning time allocation (allocate more learning time to weak knowledge points and less learning time to knowledge points that are well mastered), exercise training plan (match exercises of corresponding difficulty to weak knowledge points, set the number of training sessions and assessment standards), and clarify learning objectives and learning method suggestions.

[0031] 4) Resource Matching Unit: Based on personalized teaching plans and students' learning preferences, matching corresponding learning resources from the teaching resource library; for example, for students who prefer micro-lesson learning, pushing micro-lesson videos related to weak knowledge points; for students who prefer exercise training, pushing targeted exercise sets; for students with strong self-learning abilities, pushing supplementary materials; ensuring that the difficulty of resources is adapted to students' knowledge level (basic resources are pushed for weak knowledge points, and advanced resources are pushed for knowledge points that are well mastered), and that the type of resources matches students' learning preferences.

[0032] 4. Learning Situation Dynamic Analysis Module This module is used to monitor students' learning status in real time, accurately analyze learning progress, identify learning abnormalities and trigger interventions, providing a basis for teachers to adjust their teaching and for students to improve their learning. It specifically includes four sub-units: 1) Real-time data acquisition unit: Collects students' behavioral data (learning duration, learning frequency, resource access records) and answer data (answer results, answer time, wrong answer records, repeated wrong answer situations) in real time during the learning process. The collection frequency is no less than once per minute to ensure the real-time nature of the data and provide timely data support for learning analysis.

[0033] 2) Learning Situation Calculation Unit: Based on the knowledge graph and real-time collected data, calculate the mastery level of each student for each knowledge point (quantified using a score of 0-100, the calculation formula is: Mastery Level = (Number of Correct Answers / Total Number of Answers) × 60 + (Learning Time / Standard Learning Time) × 40). At the same time, calculate the student's learning progress (Actual Number of Knowledge Points Learned / Planned Number of Knowledge Points Learned × 100%), answer accuracy rate (Number of Correct Answers / Total Number of Answers × 100%), and error repetition rate (Number of Repeatedly Wrong Questions / Total Number of Wrong Questions × 100%) and other learning indicators.

[0034] 3) Anomaly Detection Unit: Learning progress indicator thresholds are set. When a student's learning progress indicator exceeds the threshold range, it is judged as a learning anomaly, triggering an intervention prompt. Specific threshold settings are as follows: learning progress threshold is ±10% of the corresponding grade level's teaching progress; knowledge point mastery threshold is 60 points; and answer accuracy threshold is 50%. When a student's learning progress is lower than 90% of the teaching progress, knowledge point mastery is lower than 60 points, or answer accuracy is lower than 50%, it is judged as a learning anomaly. Intervention prompts include SMS reminders and system message pushes, sent to both the teacher's and student's ends, reminding teachers to intervene promptly and students to adjust their learning plans accordingly.

[0035] 4) Report Generation Unit: Regularly generates learning analysis reports, with generation cycles set to daily, weekly, or monthly. Individual student learning reports include an overall picture of individual knowledge mastery (mastery level of each knowledge point, knowledge mastery ranking), weak knowledge points and weak knowledge chains, and learning suggestions (targeted learning methods and resource recommendations). Class-wide learning reports include the class average mastery level, common weak knowledge points (knowledge points where more than 30% of students in the class have weak mastery), and teaching optimization suggestions (adjustments to teaching methods and resource supplementation suggestions for common weak knowledge points). Learning analysis reports can be exported and printed for easy viewing by teachers and students.

[0036] 5. Intelligent Interaction Module This system is designed to enable two-way interaction between teachers, students, and the system, building a bridge for teacher-student interaction and enhancing the teaching and learning experience. Specific functions include: 1) Teacher-side functions: Teaching plan editing (viewing and modifying personalized teaching plans generated by the system, and manually adjusting the list of learning resources pushed), learning situation viewing (viewing the learning situation analysis reports of individual students and the class as a whole, and learning abnormality prompts), teaching adjustment (adjusting the teaching plan and supplementing teaching resources according to the learning situation report), online Q&A (receiving students' questions and answering them online), homework assignment and grading (assigning homework, viewing homework completion status, and grading homework online); 2) Student-side functions: Viewing learning resources (viewing personalized learning resources pushed by the system and searching for learning resources independently), answering exercises (completing exercises online, viewing answer explanations, and compiling a notebook of wrong answers), submitting questions (submitting questions encountered during the learning process to the teacher), and viewing learning progress (viewing personal learning progress analysis reports, learning anomaly alerts, and learning suggestions); 3) Online interaction between teachers and students: Supports online discussions and real-time communication between teachers and students, and among students themselves. Teachers can initiate interactive topics in class, and students can participate in discussions and share their learning experiences, thereby enhancing classroom activity and learning enthusiasm.

[0037] 6. Access Control Module This system is used for hierarchical access control of system users to ensure system data security and operational standards. System users are divided into three categories: administrators, teachers, and students, with specific permissions as follows: 1) Administrator: Has full system operation permissions, including user management (adding, deleting, and modifying user information), permission allocation (assigning corresponding permissions to teachers and students), system settings (setting data collection frequency, knowledge graph update cycle, and learning indicator thresholds), data management (viewing, backing up, and deleting all system data), and system maintenance (monitoring system operation status and handling system failures); 2) Teachers: They have access to view teaching data, edit teaching plans, intervene in student learning, assign and grade homework, and answer questions online for the corresponding class. They can only view and operate the relevant data of the class they are responsible for and cannot access the data of other classes. 3) Students: They only have the right to view their own learning data, access learning resources, submit questions, and answer exercises. They cannot view the learning data of other students or the teaching management data of teachers.

[0038] Meanwhile, this module supports dynamic adjustment of permissions. Administrators can modify the permission scope of teachers and students according to actual needs to ensure the rationality and flexibility of permission allocation.

[0039] 7. Data storage module Used to store all data during system operation, employing a distributed storage architecture (such as Hadoop distributed storage) to ensure data security, sufficient storage capacity, and efficient access; specific storage content includes: 1) All raw and preprocessed data collected by the multi-source data acquisition module; 2) Knowledge graph data generated by the knowledge graph construction and update module (initial knowledge graph, iteratively updated knowledge graph, and historical version data of knowledge graph); 3) Teaching plan data and learning resource recommendation list data generated by the personalized teaching recommendation module; 4) Learning analysis report data and learning anomaly alert data generated by the learning dynamic analysis module; 5) All log data during system operation (user operation log, data acquisition log, system fault log).

[0040] This module employs data encryption technology to encrypt and store sensitive data (such as student personal information and teacher teaching evaluation data) to prevent data leakage. It also supports data backup and recovery functions, regularly backing up data to ensure rapid recovery in case of data loss and guaranteeing stable system operation.

[0041] A knowledge graph-based smart education method The method, based on the aforementioned smart education system, includes the following six steps, forming a closed-loop teaching ecosystem of "data collection - graph construction - teaching delivery - learning analysis - graph update - ecosystem optimization". The specific steps are as follows: Step 1: Knowledge Graph Construction The knowledge extraction unit of the knowledge graph construction and update module extracts knowledge point entities, attributes, and relationships from teaching resource data collected by the multi-source data acquisition module using the BERT entity extraction model and the CNN relation extraction model, ensuring that the extraction accuracy meets the requirements. The knowledge fusion unit uses a semantic similarity-based deduplication algorithm (semantic similarity threshold 0.95) to deduplicatize and standardize the extracted knowledge, resolving knowledge redundancy and conflict issues, and mapping knowledge from different formats and sources to the same knowledge system. The knowledge storage unit uses a graph database to store the processed knowledge and constructs an initial knowledge graph. This initial knowledge graph includes knowledge point nodes, related edges, attributes, and initial student knowledge mastery status nodes (all initial mastery values ​​are set to 0, and will be updated subsequently based on student learning data).

[0042] Step 2: Multi-source data acquisition and preprocessing The system uses a multi-source data acquisition module with multiple interfaces to collect teacher teaching data, student learning data, teaching resource data, and environmental data at a frequency of no less than once per minute. The collected data undergoes preprocessing, which includes three sub-steps: S2.1: Data cleaning, removing missing values, outliers and duplicate data. Data with more than 5% missing values ​​are directly removed. Data with ≤5% missing values ​​are filled with mean or interpolation (mean is used for numerical data and interpolation is used for text data). S2.2: Data standardization converts numerical data of different magnitudes (such as learning time and test scores) into standardized data in the range of 0-1, and converts text data (such as learning preferences and teaching evaluations) into a recognizable encoding format to ensure data consistency and computability. S2.3: Data classification and storage. The pre-processed teacher teaching data, student learning data, teaching resource data, and environmental data are stored in the corresponding partitions of the data storage module, which facilitates quick query and retrieval in the future and provides data support for subsequent teaching push and learning analysis.

[0043] Step 3: Generation of Personalized Teaching Strategies The personalized teaching delivery module's student profile building unit constructs a profile of students based on preprocessed learning data, including learning preferences, knowledge mastery levels, learning abilities, and learning habits, and uses a quantitative scoring method to label the feature values ​​of each dimension. The weakness identification unit, combined with the initial knowledge graph, compares the standard value (60 points) of knowledge point mastery with the student's actual mastery level to identify the student's weak knowledge points, while also tracing the prerequisite knowledge points of these weak knowledge points to form a weak knowledge chain. The teaching plan generation unit generates personalized learning paths and teaching plans based on the student profile and the weak knowledge chain, clarifying the learning order of knowledge points, the allocation of learning time, the exercise training plan, and the learning objectives. The resource matching unit matches corresponding learning resources from the teaching resource library based on the personalized teaching plan and student learning preferences, pushes them to the student's end, and simultaneously synchronizes the personalized teaching plan to the teacher's end for viewing and adjustment.

[0044] Step 4: Dynamic Analysis and Intervention of Student Learning The learning progress analysis module uses a real-time data acquisition unit to collect student behavior and answer data during the learning process, with a collection frequency of at least once per minute. The learning progress calculation unit, based on the knowledge graph and the real-time data, calculates learning progress indicators such as the student's mastery level for each knowledge point, learning progress, answer accuracy rate, and error repetition rate. The mastery level is calculated using the formula: Mastery Level = (Number of Correct Answers / Total Number of Answers) × 60 + (Learning Time / Standard Learning Time) × 40. The anomaly detection unit compares the results against learning progress indicator thresholds (the learning progress threshold is ±10% of the corresponding grade level's teaching progress). The learning progress threshold is 60 points, and the accuracy threshold is 50%. When a student's learning indicators exceed the threshold range, it is judged as a learning abnormality, triggering an intervention prompt, which is pushed to both the teacher's and student's ends. Through the report generation unit, learning progress analysis reports for individual students and the class as a whole are generated regularly, with the generation cycle set to daily, weekly, or monthly. Teachers can view the learning progress reports through the intelligent interaction module, adjust teaching plans and personalized push strategies based on students' learning abnormalities and weaknesses, and students can conduct targeted learning based on the learning progress reports and pushed resources, complete exercises and consolidate knowledge points, and submit questions to teachers to get online answers.

[0045] Step 5: Iterative Update of Knowledge Graph The knowledge graph construction and update module's graph update unit collects real-time teaching data (teacher-added teaching resources and teaching adjustment records) and student learning data (latest answer data, learning time, and knowledge mastery level) according to a preset update cycle (default 3 days, manually adjustable). It analyzes changes in knowledge point relationships, adjustments to knowledge point attributes, and updates to student knowledge mastery status to iteratively optimize the knowledge graph. Specific updates include: adding knowledge point nodes and associated edges corresponding to new teaching resources, adjusting the difficulty and importance of knowledge points, updating student knowledge mastery status nodes, and optimizing knowledge point relationships. Simultaneously, it stores historical version data of the knowledge graph, facilitating retrospective queries of the knowledge graph's status at different times and providing historical data support for teaching optimization.

[0046] Step Six: Optimizing the Teaching Ecosystem Based on learning analysis reports and updated knowledge graph data, teachers can optimize teaching plans and resources through the intelligent interaction module, supplement teaching content for common weak knowledge points, adjust teaching methods, and improve the relevance of teaching. Students can adjust their learning plans based on personalized push notifications and learning feedback, focusing on overcoming weak knowledge points and developing good learning habits. Administrators can maintain system operation, monitor data security, adjust system settings (such as data collection frequency and learning indicator thresholds), and handle system failures through the permission management module. The system accumulates teaching and learning data through the data storage module, continuously optimizes personalized teaching push algorithms and learning analysis models, improves the relationships in the knowledge graph, and forms a closed-loop teaching ecosystem of "data collection - graph construction - teaching push - learning analysis - graph update - ecosystem optimization," continuously improving teaching efficiency and learning outcomes.

[0047] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0048] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A knowledge graph-based intelligent education system, characterized in that, include: The multi-source data acquisition module is used to collect various types of data during the teaching process, including teacher teaching data, student learning data, teaching resource data, and environmental data. The teacher teaching data includes lesson plans, courseware, teaching videos, and teaching evaluation records. The student learning data includes learning time, answer data, wrong answer records, learning preferences, and classroom interaction data. The teaching resource data includes textbook knowledge points, test questions, micro-lessons, and supplementary materials. The environmental data includes classroom atmosphere data and equipment operation data. The knowledge graph construction and update module is used to construct an initial knowledge graph based on teaching resource data and student learning data collected by the multi-source data acquisition module, and dynamically update the knowledge graph according to real-time data. The initial knowledge graph includes knowledge point nodes, knowledge point association edges, knowledge point attributes, and student knowledge mastery status nodes. The knowledge point association edges include prerequisite dependencies, parallel relationships, and extension relationships. The knowledge point attributes include difficulty coefficient, importance, and appropriate learning stage. The student knowledge mastery status nodes are associated with corresponding knowledge point nodes and labeled with mastery level values. The personalized teaching recommendation module is used to analyze students' knowledge weaknesses and learning preferences based on the knowledge graph generated by the knowledge graph construction and update module, combined with student learning data, to generate personalized teaching plans and learning resource recommendation lists. The personalized teaching plans include the order of knowledge point learning, the allocation of learning time, and the exercise training plan. The learning resource recommendation list matches corresponding micro-lessons, exercises, and extension materials according to students' weak knowledge points and learning preferences. The learning dynamic analysis module is used to collect student learning data in real time, combine it with knowledge graphs, analyze students' knowledge mastery progress, the evolution of weak knowledge points, and learning behavior patterns, generate learning analysis reports, identify student learning abnormalities and trigger intervention prompts. The learning analysis reports include individual student learning reports and class-wide learning reports. The learning abnormalities include lagging learning progress, persistently low mastery of knowledge points, and abnormal learning behaviors. The intelligent interaction module is used to realize two-way interaction between teachers, students and the system. It includes functions such as editing teaching plans, viewing student learning status and adjusting teaching on the teacher's side, viewing learning resources, answering exercises and submitting questions on the student's side, as well as online interaction and Q&A between teachers and students. The permission management module is used to manage the permissions of system users in a hierarchical manner. The users include administrators, teachers, and students. Administrators have full system operation permissions, teachers have the permissions to view teaching data, edit teaching plans, and intervene in learning progress for their corresponding classes, and students only have the permissions to view personal learning data, access learning resources, and submit questions. The data storage module stores all data collected by the multi-source data acquisition module, knowledge graph data generated by the knowledge graph construction and update module, teaching plan data generated by the personalized teaching push module, learning progress report data generated by the learning progress dynamic analysis module, and all log data during system operation. It adopts a distributed storage architecture to ensure data security and efficient access.

2. The knowledge graph-based smart education system according to claim 1, characterized in that, The knowledge graph construction and updating module includes a knowledge extraction unit, a knowledge fusion unit, a knowledge storage unit, and a graph updating unit; The knowledge extraction unit is used to extract knowledge point entities, knowledge point attributes, and knowledge point relationships from teaching resource data. It employs deep learning-based entity extraction and relationship extraction algorithms to ensure an extraction accuracy of no less than 98%. The knowledge fusion unit is used to deduplicatize and standardize the extracted knowledge point entities, attributes and relationships, solve the problem of knowledge redundancy and conflict in multi-source data, and map knowledge of different formats and sources to the same knowledge system. The knowledge storage unit is used to store knowledge graph data using a graph database, supporting quick querying and modification of knowledge point nodes and related edges, and storing historical version data of the knowledge graph for easy backtracking. The knowledge graph update unit is used to periodically update the knowledge graph based on real-time collected student learning data, teacher teaching data, and newly added teaching resource data. The update cycle can be set to 1-7 days, and the update content includes adjusting the relationship between knowledge points, updating the attributes of knowledge points, and updating the student knowledge mastery status nodes.

3. The knowledge graph-based smart education system according to claim 1, characterized in that, The personalized teaching push module includes a student profile building unit, a weakness identification unit, a teaching plan generation unit, and a resource matching unit; The student profile building unit is used to build a student profile based on student learning data, including learning preferences, knowledge mastery level, learning ability, and learning habits, and to label the feature values ​​of each dimension using a quantitative scoring method. The weak point identification unit is used to combine knowledge graph and student learning data, and identify the student's weak knowledge points by comparing the standard value of the knowledge point mastery with the student's actual mastery value. At the same time, it analyzes the pre-dependent knowledge points of the weak knowledge points to form a weak knowledge chain. The teaching plan generation unit is used to generate personalized learning paths and teaching plans based on student profiles and weak knowledge chains, specifying the learning time, learning methods and assessment standards for each knowledge point; The resource matching unit is used to match corresponding learning resources from the teaching resource library according to the personalized teaching plan and students' learning preferences, so as to ensure that the difficulty of the resources is suitable for the students' knowledge level and the type of resources is matched with the students' learning preferences.

4. The knowledge graph-based smart education system according to claim 1, characterized in that, The learning dynamic analysis module includes a real-time data acquisition unit, a learning calculation unit, an anomaly identification unit, and a report generation unit; The real-time data acquisition unit is used to collect students' learning behavior data and answer data in real time, with a collection frequency of no less than once per minute to ensure the real-time nature of the data. The learning progress calculation unit is used to calculate the mastery level value of each knowledge point of the student based on the knowledge graph. The mastery level value is quantified from 0 to 100 points. At the same time, it calculates the student's learning progress, answer accuracy rate, and error repetition rate, among other learning progress indicators. The anomaly identification unit is used to set learning indicator thresholds. When a student's learning indicator exceeds the threshold range, it is determined to be a learning anomaly, triggering an intervention prompt. The threshold can be adjusted by the teacher according to the actual situation of the class. The report generation unit is used to periodically generate learning analysis reports. A single student's learning report includes an overview of their knowledge mastery, weak knowledge points, and learning suggestions. A class-wide learning report includes the class's average mastery level, common weak knowledge points, and teaching optimization suggestions.

5. A knowledge graph-based smart education method, characterized in that, Includes the following steps: Step 1: Knowledge graph construction. Through the knowledge extraction unit of the knowledge graph construction and update module, knowledge point entities, knowledge point attributes, and knowledge point relationships are extracted from the teaching resource data collected by the multi-source data acquisition module. The extracted knowledge is deduplicated and standardized by the knowledge fusion unit to resolve knowledge redundancy and conflict. The knowledge is stored in a graph database through a knowledge storage unit to construct an initial knowledge graph. The initial knowledge graph includes knowledge point nodes, related edges, attributes, and initial student knowledge mastery status nodes. Step 2: Multi-source data acquisition and preprocessing. Through the multi-source data acquisition module, teacher teaching data, student learning data, teaching resource data, and environmental data are collected. The collected data is cleaned, deduplicated, and standardized to remove invalid data. Data of different formats is converted into a unified format and stored in the data storage module to provide data support for subsequent teaching push and learning analysis. Step 3: Personalized teaching strategy generation. The student profile building unit in the personalized teaching push module constructs student profiles based on preprocessed student learning data; the weakness identification unit identifies students' weak knowledge points and weak knowledge chains by combining the initial knowledge graph; the teaching plan generation unit generates personalized teaching plans based on student profiles and weak knowledge chains; and the resource matching unit matches corresponding learning resources and pushes them to the student's end, while simultaneously synchronizing the personalized teaching plans to the teacher's end. Step Four: Dynamic Analysis and Intervention of Learning Progress. The dynamic learning progress analysis module uses a real-time data collection unit to gather students' behavioral and answer data during the learning process. The learning progress calculation unit, based on the knowledge graph and the real-time collected data, calculates learning progress indicators such as students' mastery of knowledge points and learning progress. The anomaly identification unit compares learning progress indicator thresholds to identify learning anomalies and triggers intervention prompts, which are then pushed to both teachers and students. The report generation unit regularly generates learning progress analysis reports for individual students and the entire class, providing a basis for teachers to adjust their teaching and for students to improve their learning. Teachers can view the learning progress reports through the intelligent interaction module, adjust teaching plans and personalized push strategies based on students' learning anomalies and weaknesses, and students can engage in targeted learning based on the learning progress reports and pushed resources. Step 5: Iterative update of the knowledge graph. Through the graph update unit of the knowledge graph construction and update module, real-time teaching data and student learning data are collected regularly. The changes in the relationship between knowledge points, the adjustment of knowledge point attributes, and the updates of students' knowledge mastery status are analyzed. The knowledge graph is iteratively optimized, and knowledge point nodes, related edges, attributes, and student knowledge mastery status nodes are updated to ensure that the knowledge graph is synchronized with the actual teaching. At the same time, the historical versions of the knowledge graph are stored for easy backtracking and querying. Step Six: Optimizing the Teaching Ecosystem. Based on the learning analysis report and updated knowledge graph data, teachers optimize teaching plans and resources through the intelligent interaction module, students adjust their learning plans according to personalized push notifications and learning feedback, and administrators maintain system operation and monitor data security through the permission management module. The system accumulates teaching and learning data through the data storage module, forming a closed-loop teaching ecosystem of "data collection - knowledge graph construction - teaching push - learning analysis - knowledge graph update - ecosystem optimization", continuously improving teaching efficiency and learning outcomes.

6. The knowledge graph-based smart education method according to claim 5, characterized in that, In step one, the deep learning algorithms used by the knowledge extraction unit include the BERT entity extraction model and the CNN relation extraction model, wherein the extraction accuracy of the BERT entity extraction model is no less than 98%, and the extraction accuracy of the CNN relation extraction model is no less than 97%. The knowledge fusion unit adopts a deduplication algorithm based on semantic similarity. The semantic similarity threshold is set to 0.

95. When the semantic similarity of two knowledge point entities is ≥0.95, they are determined to be the same entity and are merged.

7. The knowledge graph-based smart education method according to claim 5, characterized in that, Step two, data preprocessing includes the following sub-steps: S2.1: Data cleaning, removing missing values, outliers and duplicate data. Data with more than 5% missing values ​​are directly removed, while data with ≤5% missing values ​​are filled with mean or interpolation. S2.2: Data standardization, converting numerical data of different magnitudes into standardized data in the 0-1 range, and converting text data into a recognizable encoding format to ensure data consistency; S2.3: Data is classified and stored separately. The pre-processed teacher teaching data, student learning data, teaching resource data and environmental data are stored in the corresponding partitions of the data storage module for easy retrieval and retrieval later.

8. The knowledge graph-based smart education method according to claim 5, characterized in that, In step four, the learning indicators include learning progress threshold, knowledge point mastery threshold, and answer accuracy threshold. The learning progress threshold is set to ±10% of the teaching progress of the corresponding grade level, the knowledge point mastery threshold is set to 60 points, and the answer accuracy threshold is set to 50%. When a student's learning progress is lower than 90% of the teaching progress, the knowledge point mastery is lower than 60 points, or the answer accuracy is lower than 50%, it is judged as an abnormal learning situation, triggering an intervention prompt.

9. The knowledge graph-based smart education method according to claim 5, characterized in that, In step five, the update cycle of the knowledge graph can be set by the administrator or the teacher, with a default update cycle of 3 days. The update content includes: adding knowledge point nodes and related edges corresponding to the new teaching resources, adjusting the difficulty coefficient and importance of knowledge points, updating student knowledge mastery status nodes, and optimizing the relationship between knowledge points. Among them, student knowledge mastery status nodes are updated in real time based on the latest student answer data and learning time.

10. A knowledge graph-based smart education method according to claim 5, characterized in that, Step six, teaching ecosystem optimization also includes system performance optimization, which improves data access speed through the distributed storage architecture of the data storage module and ensures that the system's concurrent access volume is no less than 1,000 people online at the same time; The system ensures data security by regularly updating user permissions through the permission management module and improves user experience by optimizing the user interface through the intelligent interaction module.

Images