Multifunctional integrated self-service printing learning machine

Abstract

The invention relates to a multifunctional integrated self-service printing learning machine, and aims to provide comprehensive learning assistance from question selection, paper answering, automatic correction to wrong question management and personalized learning report generation for students. The learning machine comprises a printer main body, a touch display screen, a paper feeding clamping groove and a printing outlet. The equipment realizes efficient data transmission by embedding a database and a near data processing (NDP) module, and supports duplex serial communication. The system combines artificial intelligence technologies such as NLP (Natural Language Processing), similarity analysis, GNN (Graph Neural Network) and the like to intelligently analyze answers of students. The device can automatically generate error records and personalized notes, help students to identify and review weak knowledge points, generate knowledge maps and learning reports, and provide personalized learning suggestions for the students. The system is low in cost, is easy to operate, is suitable for being popularized and used in educational institutions, and provides efficient and convenient self-service learning service for students.

Description

Technical Field

[0001] The present invention relates to the technical field of learning machines, and in particular to a multifunctional integrated self-service printing learning machine. Background Art

[0002] With the development of educational informatization, various types of educational auxiliary equipment have appeared on the market, mainly concentrated in tablet-type learning machines. This type of learning machine has strong learning functions, but due to the high cost of the equipment and low popularity, it is difficult for more families, especially middle-class and lower-income students, to enjoy high-quality educational resources. At the same time, traditional printers on the market are mainly used for document printing and scanning, and cannot bear the educational functions of learning machines. The functions of learning machines and printers are usually separated and cannot be integrated, resulting in students needing to connect multiple devices during use, which is inconvenient and increases costs.

[0003] Although existing tablet learning machines can meet some of the needs of self-learning, they mainly rely on electronic screens to display content. It is difficult for students to answer questions and sort out wrong questions on paper media, and they cannot meet the needs of most learning scenarios that mainly rely on paper answers. In addition, students need to spend a lot of time sorting out and analyzing wrong questions, and many people cannot effectively identify and mark weak knowledge points during the process of sorting out wrong questions, which affects the learning effect. Although there are some auxiliary software on the market that can manage wrong questions, most of them are limited to personal use, lack in-depth analysis of students' personalized learning characteristics, and are difficult to apply to large-scale personalized learning services.

[0004] Based on this, how to effectively integrate the functions of learning machines and printers, and introduce artificial intelligence technology to provide an intelligent device that integrates personalized learning, question selection, automatic correction, wrong question sorting, career planning and other functions has become an urgent problem to be solved in the field of education. The device should have the following features: support paper answers and feedback, intelligent diagnosis of students' weak knowledge points, automatic generation of personalized learning reports, and prediction of students' future career adaptation to improve students' learning efficiency and reduce education costs. This integrated self-service learning device enabled by artificial intelligence will make up for the shortcomings of existing educational products, enable students to obtain personalized learning guidance more conveniently, and has broad application prospects. Summary of the invention

[0005] In response to the above problems, the present invention provides a multifunctional integrated self-service printing learning machine that integrates artificial intelligence technology. The learning machine realizes the integration of printer and learning terminal at the hardware level, and provides functions such as question printing, paper answering, answer sheet scanning, etc.; at the software level, it introduces artificial intelligence technologies such as knowledge graph, big data analysis, natural language processing, etc., to realize personalized learning analysis, intelligent diagnosis of wrong questions, career planning prediction and other intelligent functions, aiming to provide students with convenient, economical and efficient intelligent learning tools. The main technical solutions are as follows:

[0006] Multifunctional integrated hardware design

[0007] Printer body: the core part of the device, integrating printing, scanning, storage, display and other functions. The front end is equipped with a touch screen for students to interact; the lower part is equipped with a paper feed slot, which supports automatic paper feeding and continuous printing; the front end is equipped with a printing outlet to output paper materials such as test papers and study reports.

[0008] Smart terminal: Embedded with Android / Linux operating system, running learning software and artificial intelligence algorithms. The terminal communicates with the printer via USB or wired / wireless network to control printing tasks and collect student answer data. The terminal is equipped with a capacitive touch screen, high-definition camera, microphone array, speaker, etc., supporting multi-modal human-computer interaction.

[0009] Personalized learning analytics

[0010] Knowledge graph construction: Based on textbooks and question banks, knowledge points are extracted and relationships are constructed to form a structured knowledge graph. Knowledge points are mapped to questions through automated tools and expert annotations.

[0011] Answer data collection: After students answer the questions on the paper answer sheet, they insert the answer sheet into the paper card slot at the bottom of the printer. The printer scans the answer sheet through the high-definition scanning module, collects the answer data and uploads it to the smart terminal.

[0012] Personalized diagnosis and analysis: Based on the students’ answer data, the intelligent terminal uses the AILBD algorithm to compare the knowledge graph and diagnose the students’ mastery of knowledge points. Through wrong question clustering and associative reasoning, it analyzes the students’ weak knowledge points and frequent error patterns and generates personalized learning reports.

[0013] Personalized learning resource push: Based on the diagnosis results, the system automatically recommends micro-class videos, exercises, etc. related to weak knowledge points to guide students to fill in the gaps; based on the students' knowledge map, customized exercises are provided.

[0014] Wrong question collection and intelligent analysis

[0015] Wrong question collection: After students answer, the intelligent terminal compares the answer sheet with the standard answer, automatically scores and collects wrong question data. It supports students to actively mark difficult questions and doubts.

[0016] Automatic association of wrong question knowledge points: Based on OCR and NLP technology, the text features of wrong questions are extracted, and the knowledge points tested in the questions are automatically associated through the knowledge graph, and connections are established with students’ weak knowledge points.

[0017] Intelligent analysis of error causes: Through image recognition and semantic analysis technology, the problem-solving process and answering logic of students in wrong questions are analyzed to infer possible causes of errors, such as unclear question review, calculation errors, confusion of concepts, etc.

[0018] Intelligent generation of wrong questions book: The system automatically collects wrong questions into students' personal wrong questions book, and generates wrong questions knowledge cards based on the knowledge graph, giving explanations of relevant knowledge points and recommendations of similar questions for each wrong question. It supports cross-terminal synchronization of wrong questions book and wrong questions management.

[0019] Career Planning Intelligent Prediction

[0020] Analysis of interests and advantageous subjects: The Holland Career Test is used to obtain the characteristics of students’ career interests; based on the Z-score, T-score, and percentile calculations of students’ grades in each subject, combined with the interquartile range, the top three advantageous subjects of students are obtained.

[0021] Decision tree model training: Use social personnel data to train the decision tree, where the input features are dominant disciplines and Holland's six types of scores, and the target variable is occupation type.

[0022] Prediction and recommendation of suitable careers: Input students’ advantageous subjects and career interest characteristics into the decision tree model to predict their suitable career types in the future and recommend related careers. Combined with the ideal careers chosen by students, a conclusion is given as to whether they are suitable.

[0023] Mental Health Assessment and Intervention

[0024] Psychological assessment: Regularly push personalized psychological test questions, analyze students' psychological state through scales, and form a mental health file.

[0025] Analysis of evaluation results: Use correlation analysis, data mining and other methods to explore the relationship between subject mastery and psychological state, and predict the trend of students' psychological state changes in the future.

[0026] Psychological intervention: Based on the analysis results, personalized psychological health guidance content is pushed to relieve students' learning pressure. For students with prominent problems, timely warnings are provided and professional psychological counseling channels are provided.

[0027] Photo search and analysis

[0028] Question image recognition: Students take pictures of questions through the camera of the smart terminal, and the system uses OCR technology to perform text recognition.

[0029] Question matching and analysis: Extract key information from the question stem and match it with the question bank. If it cannot be matched directly, use NLP technology to analyze the question type and knowledge points, and give similar questions and solution ideas.

[0030] Related display of problem-solving process: For the knowledge points related to the questions, the system obtains the mastery status from the students' knowledge map and provides targeted reinforcement and prompts in the problem-solving ideas.

[0031] Application of subject knowledge graph

[0032] Graph personalization: Based on students’ learning data, their personal knowledge graph is dynamically updated to reflect their knowledge mastery in real time.

[0033] Graph reasoning: Use the relational reasoning ability of graphs to predict the learning difficulties that students may encounter and provide pre-resources and learning guidance.

[0034] Knowledge tracing: When students encounter problems, they can trace back to the source concepts that caused the problems from the knowledge graph and supplement basic knowledge in a targeted manner.

[0035] The present invention integrates a variety of artificial intelligence technologies and deeply combines them with educational scenarios, greatly improving the intelligence level of the learning machine. Functions such as personalized learning analysis, wrong question diagnosis, and career planning prediction can accurately analyze the learning situation, provide students with personalized learning guidance, and improve learning efficiency. The multifunctional integrated design reduces the cost of equipment and use, so that intelligent learning services can benefit more student groups. The application of this learning machine will promote teaching students in accordance with their aptitude and maximize the learning potential of individual students.

[0036] Advantageous Effects of the Invention

[0037] The multifunctional integrated self-service printing learning machine provided by the present invention has the following significant advantages:

[0038] Cost-effectiveness: The device has built-in printing, scanning and learning analysis modules, which realizes the integration of learning and printing and reduces the cost of equipment. Compared with traditional learning machines (tablet type), it eliminates the need for external printing equipment. One device can meet the needs of multiple students and is suitable for large-scale application in educational institutions.

[0039] Efficient and convenient: The device automatically completes question selection, correction, wrong question management and knowledge point marking, reducing the workload of teachers and improving students' learning efficiency. After students finish answering, they can immediately get feedback and wrong question analysis, shortening the waiting time.

[0040] Personalized learning support: The system automatically generates personalized reports on wrong questions and weak knowledge points based on students’ learning situation, helping students focus on reviewing weak points and improving learning efficiency.

[0041] Accurate feedback and intelligent analysis: The device integrates NLP, similarity analysis and clustering computing technologies to intelligently analyze students' answers to ensure the accuracy of feedback. Wrong question records and knowledge graphs help students sort out learning content and gradually form a knowledge system.

[0042] Through the implementation of the present invention, students can obtain integrated learning services, including question selection, paper answers, automatic correction, wrong question sorting and generation of personalized learning reports, thereby significantly improving learning effects and saving learning time. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0044] Figure 1 A schematic diagram of the overall hardware structure of a learning machine according to an embodiment of the present invention;

[0045] Figure 2 A schematic diagram of the working principle structure of a learning machine according to an embodiment of the present invention;

[0046] Figure 3 This is a schematic diagram of the web page operation interface of the learning machine according to an embodiment of the present invention;

[0047] Figure 4 This is a schematic diagram of the web page operation interface of the learning machine according to an embodiment of the present invention;

[0048] Figure 5 A schematic diagram of a basic process of generating a decision tree according to an embodiment of the present invention.

[0049] The markings in the figure are:

[0050] 1. Paper feed slot; 2. Display screen; 3. Test paper entrance; 4. Print exit. DETAILED DESCRIPTION

[0051] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0052] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0053] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0054] The present invention provides a multifunctional integrated self-service printing learning machine, which integrates the learning machine with the printer function to achieve full-process learning assistance from question selection, paper answers, automatic correction to wrong question management, personalized note generation, etc. The device design and functional layout enable it to provide students with a convenient personalized learning experience. The following is a specific implementation of the present invention:

[0055] See also Figures 1 to 5

[0056] Multifunctional integrated self-service printing learning machine

[0057] Example 1: Integrated hardware design

[0058] Printer body: The core part of the device, with functions such as printing, scanning, storage, and display. The front end is equipped with a 7-inch capacitive touch screen with a resolution of 1024*600 and supports multi-touch, which is convenient for students to select questions, view analysis, and personalized learning management. There is a paper feed slot at the bottom of the front end of the device, which supports continuous automatic paper feeding of A4 format, with a maximum load of 100 pages, meeting the needs of students to print test papers and reports in batches. The print outlet is located on the right side of the machine body, using thermal printing and inkjet printing dual engines, supporting black and white and color output. The printing resolution is 600dpi for black and white and 1200dpi for color.

[0059] Smart terminal: Equipped with a quad-core 1.8GHz ARM processor, 4GB RAM + 64GB storage space, and pre-installed Android 11 operating system. The terminal is connected to the printer via a MicroUSB interface and supports 2.4G / 5G dual-band WiFi wireless communication. It is equipped with an 8-megapixel autofocus camera for scanning answer sheets; the microphone and speaker support voice interaction. The display is a 10.1-inch fully laminated IPS capacitive screen with a resolution of 1920*1200. The terminal has a built-in 5000mAh lithium battery to support long-term mobile use.

[0060] Example 2: Personalized learning analysis

[0061] Knowledge graph construction: Extract knowledge topics and their internal connections from multi-source heterogeneous data such as textbooks, question banks, and dictionaries, and use a bottom-up hierarchical clustering algorithm combined with manual annotation to build a large-scale knowledge graph covering 12 subjects in junior high and high school. The graph scale exceeds 100,000 knowledge nodes and 1 million knowledge relationships. The mapping of knowledge points and exercises uses a combination of rule-based automatic association and manual annotation to ensure the accuracy of the association.

[0062] Answer data collection: Supports two answer sheet formats: OMR (optical mark recognition) and OCR (optical character recognition). For multiple-choice questions, students fill in the answers in the preset OMR dot area, and each answer sheet can recognize 200 options; for subjective questions, students write their answers in the OCR area, and the smart terminal directly recognizes the handwritten content through OCR technology. The answer data is automatically uploaded to the cloud for storage.

[0063] Personalized diagnosis and analysis: Based on the students' answer data, the intelligent terminal uses the improved AILBD (answer knowledge point association-wrong question diagnosis-knowledge point relationship deduction-learning situation reasoning decision) algorithm to accurately depict the students' knowledge map. Through the combination of deep expert knowledge and data-driven, the students' strengths and weaknesses are comprehensively analyzed in multiple dimensions. Multi-factor weighted scoring is used to scientifically quantify the degree of knowledge point mastery. Based on graph embedding representation learning, the association patterns and knowledge gaps between wrong questions are mined. The causes of errors are revealed through causal inference, and hidden weak knowledge points are predicted.

[0064] Personalized learning resource push: Comprehensive historical learning performance and diagnostic results to form a multi-label learning portrait of students. Based on collaborative filtering and transfer learning algorithms, personalized recommendations are made from multi-source heterogeneous resources such as question banks, micro-courses, and test papers. According to students' cognitive level and subject preferences, personalized test papers with increasing difficulty and comprehensive question types are customized. The recommended resources are highly correlated with weak knowledge points, helping students to weaken their weaknesses and strengthen their strengths.

[0065] Example 3: Wrong Question Collection and Intelligent Analysis

[0066] Wrong question collection: Compare students' answer sheets with standard answers, automatically grade and extract wrong questions. Supports filtering wrong questions by question type, difficulty, knowledge points and other dimensions. Students can mark difficult questions and doubts by voice or handwritten annotations.

[0067] Automatic association of wrong questions with knowledge points: Integrate multi-strategy algorithms based on rules, statistics, and semantics to extract key phrase features of wrong questions and map them to one or more knowledge points in the knowledge graph. Continuously iterate and optimize the granularity of knowledge points and association thresholds to balance the accuracy of granularity and association. Introduce the attention mechanism to give high weights to key fragments of the question stem, options, and analysis to improve the association effect.

[0068] Intelligent analysis of error causes: Build a digital record of the entire process of students answering questions, and restore the problem-solving ideas through multimodal data analysis such as eye movement, handwriting, and voice. Automatically diagnose the causes of errors by comparing with standard ideas, covering more than 20 dimensions of error causes such as question review, conception, concept, formula, and calculation. Use causal reasoning to reveal the logical chain between error causes. Based on the clustering of similar wrong questions, explore the typical error patterns of students. Form structured annotations for error causes to support statistical analysis and personalized error correction.

[0069] Intelligent generation of wrong questions book: Automatically classify wrong questions into corresponding subjects and knowledge points to form a graded wrong question book. Intelligently mine multi-source data such as textbooks, dictionaries, and videos for the weak knowledge points associated with each wrong question, and summarize background knowledge such as concepts, theorems, and formulas. Recommend corresponding micro-class videos and error correction exercises to form a full-link review closed loop. Wrong questions can be reviewed across platforms through cloud synchronization. The wrong question book supports personalized screening by question type, difficulty, and mastery level.

[0070] Example 4: Intelligent prediction of career planning, such as Figure 5 Shown

[0071] Analysis of interests and advantageous disciplines: Through the Holland Vocational Interest Scale and professional personality test, students' professional personality, professional interests, personality traits, etc. are comprehensively examined. Based on big data analysis, the intrinsic correlation between subject performance and personality traits is discovered. The improved IV-CART (importance-variance-based classification and regression tree) algorithm is used to comprehensively evaluate subject advantages from multiple angles such as Z scores, T scores, and percentiles. Indicators such as Q statistics are used to characterize the consistency of academic performance and interests. The task completes the overall process.

[0072] Decision Tree is a model that uses a tree-like data structure to display decision rules and classification results. As an inductive learning algorithm, its focus is to transform seemingly disordered and messy known data into a tree-like model that can predict unknown data through some technical means. Each path from the root node (the attribute that contributes the most to the final classification result) to the leaf node (the final classification result) represents a decision rule. In the specific application scenario, in layman's terms, it is to quantify the multiple characteristic attributes of a thing into meaningful data forms and form vectors. Then, an item with n characteristic attributes will be converted into an n-dimensional feature vector, and some means will be taken to divide the data into different categories according to the data distribution of the sample data set. The various features of the target item can be gradually classified to achieve the final purpose of classifying the item.

[0073] Root Node In a decision tree, the root node is the starting point of the tree and the beginning of the entire decision process. This node is the highest level of the decision tree, and it determines the subsequent decision path based on the student's dominant subject. The selection of the root node is based on the analysis of the data set, and a feature that can best distinguish different results is selected as the starting point.

[0074] Leaf Node: A leaf node is the terminal node of a decision tree and does not perform further splitting. In the context of career planning prediction, a leaf node represents the final career category prediction.

[0075] Boundary Points (Split Point) Boundary points are specific values ​​used in decision trees to split a dataset. At each decision node, we select a feature and a threshold (boundary point) to divide the data into two or more subsets. The selection of these split points is based on the principle of maximizing information gain or other criteria such as Gini impurity or mean squared error.

[0076] Data Partitioning: In a decision tree, each internal node represents a feature and a boundary point, which is used to split the data into two or more subsets. This process is repeated until a leaf node is reached, which represents the final decision result.

[0077] Decision Path: The path from the root node to the leaf node represents a series of decision rules that are used to predict the career direction of students. The decision of each internal node is based on specific features and boundary points.

[0078] ① Prepare a detailed list of occupational classifications to be used as the target classification standard, which can be obtained through expert interviews or questionnaire surveys;

[0079] ②Dataset survey: The questionnaire was used to obtain the dominant disciplines, interest type scores and occupational types of people with high job satisfaction, and the decision tree training data set was obtained;

[0080] ③ Use the data set to train the decision tree for automatic classification, with the dominant disciplines and interest types as features (input variables) and the suitable occupation type as the target variable (output variable);

[0081] Decision tree model training:

[0082] Information entropy and conditional entropy (prerequisites for feature selection)

[0083] Calculate information entropy: At the root node of the decision tree, calculate the information entropy of the entire data set to evaluate the uncertainty of the target variable (occupation type) without any splitting conditions. For a discrete random variable X, its probability distribution is P(X), and the information entropy calculation is shown in the figure; where P(xi) is the probability of the random variable X taking the i-th value, b is the base of log, usually 2, and the unit of entropy is bit.

[0084]

[0085] Calculate conditional entropy: For each feature (dominant subject and interest type), calculate the conditional entropy to assess the uncertainty of the target variable given the known values ​​of these features. Conditional entropy is a measure of the uncertainty of a random variable given the known values ​​of another random variable. For random variables X and Y, the conditional entropy H(X|Y) is calculated as shown in the figure; where P(yj) is the probability that the random variable Y takes the jth value, and H(X|Y=yj) is the conditional entropy of X given that Y takes the jth value.

[0086]

[0087] Feature selection and node splitting

[0088] Information gain: For each feature, calculate the information gain, which is the weighted average of the information entropy of the parent node minus the conditional entropy of the child node after the feature is split. Features with high information gain will be preferred for node splitting.

[0089]

[0090] Among them, H(Y) is the information entropy of the target variable Y, H(Y|X=xi) is the conditional entropy of the target variable when the feature X takes the value of xi, Di is the data subset when the feature X takes the value of xi, and D is the entire data set.

[0091] Node splitting: Select the feature with the largest information gain for node splitting, and recursively construct a decision tree.

[0092] Model evaluation and optimization

[0093] Pruning: To prevent overfitting, pruning techniques can be used to limit the growth of the tree.

[0094] Cross-validation: Use cross-validation to evaluate the generalization ability of the model.

[0095] ④ Input the student’s strengths in subjects and interest scores, and use decision tree classification to predict the student’s matching career type.

[0096] Then obtain the student's career plan (if possible, perform a simple classification of possible career positions - the final goal is the same as the classification of the target variable of the decision tree), compare it with the classification result of the decision tree, and output a True or False result, that is, whether the student's current academic performance and interests match his or her current career plan.

[0097] Decision tree model training: The historical academic and career data of college students and graduates are selected as training samples. Feature selection and split point determination are optimized based on information gain ratio, Gini index and other criteria to ensure the generalization ability of classification decision trees. SMOTE, AdaBoost and other technologies are used to process unbalanced data. Model hyperparameters are tuned through cross-validation and grid search. Decision rules are introduced in combination with expert knowledge to balance the accuracy and interpretability of the model.

[0098] Prediction and recommendation of suitable careers: trigger intelligent prediction with one click through student portrait features, and return the top 10 suitable majors and occupations online. Analyze the career development path map based on the student's ideal career, and give similar career recommendations by comparing the advantageous majors. Combine the employment market situation and the employer portrait to predict career prospects and promotion space. Analyze the admission situation and employment quality of the past years for the suitable majors, and recommend the transcripts of alumni interviews.

[0099] Example 5: Mental Health Assessment and Intervention

[0100] Psychological assessment: Relying on the authoritative psychological scale library, the knowledge graph and rule engine are used to realize the intelligent generation of personalized psychological assessment questions. It covers multiple dimensions such as learning psychology, personality psychology, and social psychology. It supports accurate push and autonomous screening. The assessment forms include scales, projections, scenarios, etc. The assessment data is associated with the student portrait to form a continuous mental health time series archive.

[0101] Analysis of evaluation results: Clean the evaluation data, perform feature engineering and dimension reduction, and extract psychological state indicators. Reveal the correlation pattern between learning state and psychological state through correlation analysis, regression analysis, etc. Use time series algorithms to depict the dynamic change trajectory of psychological state and profile key psychological events. Use anomaly detection algorithms to warn of sudden changes in psychological state and extreme emotions in real time.

[0102] Psychological intervention: Benchmarking the National Student Mental Health Education Guidelines, targeted counseling content is pushed. Relying on cognitive behavioral therapy, positive psychology, etc., personalized psychological intervention programs are designed. Intervention forms include counseling information, interactive exercises, scenario simulations, etc. Introduce virtual psychological counseling assistants to provide listening 24-hour responses. Integrate high-quality social psychological counseling resources and establish green referral channels. The intervention effect evaluation and psychological state analysis form a closed loop.

[0103] Example 6: Photo search and analysis

[0104] Question image recognition: Through the ultra-large-scale OCR model with 1.3 billion parameters, the recognition accuracy exceeds 99%. 5 million question recognition templates are preset, covering various subjects, question types, and grades. Visual enhancement technology is used to optimize image quality. Supports multi-angle, paper / electronic materials and other multi-scene photo recognition. After recognition, the question stem, options, analysis, etc. can be edited and corrected.

[0105] Question matching and analysis: Match the key information of the question with the knowledge graph entity, and search for questions of the same question type and the same test point in priority. Sort and select the candidate questions based on the similarity of the question surface and the coverage of knowledge points. For new questions, obtain the answer rules of similar questions based on transfer learning and generate personalized problem-solving ideas. Ideas give priority to recommending knowledge points that students have mastered, and progressively introduce new knowledge points. Problem-solving ideas interact with the knowledge graph in real time.

[0106] Problem-solving process association display: Record key steps such as writing and drawing in the problem-solving process of students. Formalize the problem-solving process of students through handwriting recognition, image recognition and other technologies. Provide enhanced prompts and correction interventions for key knowledge points in the problem-solving process based on the students' mastery of personal knowledge graphs. Support real-time evaluation of the standardization, efficiency and logic of the problem-solving process. Reflect, summarize and consolidate standard problem-solving cases.

[0107] Example 7: Subject Knowledge Graph Application

[0108] Graph Personalization: Every time a student answers or reviews, the system automatically captures the changes in their cognitive state and adjusts the mastery weight of knowledge points in the knowledge graph in real time. Visualize the mastery distribution of the knowledge graph. For the completeness and accuracy of the personalized graph, students are supported to manually edit and correct it. The personalized knowledge graph can be accumulated and migrated to the new school year.

[0109] Graph reasoning: Based on the multi-hop relationship reasoning of the graph, predict the key and difficult knowledge of the next stage. Targeted push of pre-knowledge points sorting resources. Discover the key core knowledge in the graph, summarize and construct the main line of knowledge, and guide the systematic learning path. Use community detection and centrality analysis to reveal the macro connection of knowledge graphs of different disciplines and cultivate interdisciplinary thinking.

[0110] Knowledge tracing: Locate the weak sources in the knowledge graph by tracing back the relationship paths. Systematically supplement the basic knowledge and summarize and build the knowledge tree. For frequently forgotten or confused knowledge points, trace the source to analyze the strength, stability and logical chain of knowledge association, and take targeted measures to find out the gaps and fill them. Associate weak knowledge points with similar question types and real question situations to strengthen knowledge internalization.

[0111] The innovation of this invention is to integrate the printer with the intelligent learning terminal, empower the traditional learning scene through artificial intelligence algorithms, and provide students with personalized intelligent learning services throughout the entire process. The learning machine can accurately analyze the learning situation, explore the learning characteristics and cognitive laws of students, form personalized knowledge maps and ability portraits, and dynamically update and iterate.

[0112] Based on the traditional advantages of answering questions with pen and paper, this invention incorporates cutting-edge technologies such as OCR recognition, knowledge graph, and intelligent diagnosis of wrong questions, realizing digital collection and intelligent analysis of the learning process. Students can immediately obtain personalized diagnosis reports and learning resource push after answering, and conduct targeted review through wrong question books. The answer data and subject knowledge graphs verify each other, realizing a closed loop of knowledge internalization. This human-computer collaborative learning method improves learning efficiency, helps students to find gaps and teach students in accordance with their aptitude.

[0113] The present invention also introduces extended services such as career planning prediction and mental health analysis, making full use of academic and psychological big data, and providing comprehensive growth guidance for students through data mining and machine learning technology to understand their development. Students can make career plans early and receive continuous mental health protection.

[0114] In general, the present invention systematically designs and optimizes artificial intelligence technology for educational application scenarios, and builds an immersive smart learning ecosystem. The multifunctional integrated design reduces the terminal threshold, and personalized intelligent services are expected to allow every student to enjoy fair and quality education. The present invention reflects the development direction of artificial intelligence-enabled education and has significant social and economic value.

[0115] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A multifunctional integrated self-service printing learning machine, comprising a printer body, characterized in that The printer body has printing, scanning, storage and display functions, and can support students to independently complete question selection, answering, wrong question analysis and personalized learning management; the front end of the printer body is provided with a touch-operated display screen (2), which is used for students to select questions, view standard answers for self-assessment, personalized wrong question analysis and knowledge point report; the front lower part of the printer body is provided with a paper feed slot (1), the paper feed slot (1) has an automatic paper feeding function, supports continuous printing, and is convenient for students to select and print question content; the front end of the printer body is also provided with a printing outlet (4), the printing outlet (4) is used to output printed question papers, knowledge point analysis and wrong question summary papers, so as to support students to answer and review conveniently through paper materials and improve learning effects, and the top of the printer body is also provided with a test paper entrance (3).

2. The multifunctional integrated self-service printing learning machine according to claim 1 is characterized in that The learning machine external database management system is used to store answer sheet ID, questions, standard answers and student learning records. Students can conduct self-assessment by comparing with the standard answers provided by the system. The database can select SQL database for large-scale data storage, or select SQLite database for lightweight data storage; the learning machine embeds a near data processing (NDP) module in the printer body to convert image data into base64 format for transmission, thereby reducing the amount of data I / O, reducing latency and broadband burden, and adopts the UDP communication method of Hongmeng software, supporting RTP or UARP duplex serial protocol to ensure stable data transmission.

3. The multifunctional integrated self-service printing learning machine according to any one of claims 1 to 2 is characterized in that The learning machine has the functions of question selection and automatic correction. After the student selects the knowledge point and question content on the display screen, the device retrieves the corresponding question from the database and prints it; after completing the answer, the answer sheet is inserted into the test paper entrance (3), the device scans the answer sheet and automatically compares the standard answer to generate answer feedback; the mobile phone end and PC end of the learning machine are connected to two databases respectively, one database stores the standard answer of the question, and the other database stores the student's answer information and learning records to support personalized learning analysis.

4. The multifunctional integrated self-service printing learning machine according to any one of claims 1 to 3, characterized in that The learning machine includes a personalized learning analysis module, which constructs a subject knowledge graph, collects student answer data, uses the improved AILBD algorithm to perform personalized learning diagnosis, and generates personalized learning reports and resource push; the learning machine also includes a wrong question collection and intelligent analysis module, which automatically generates a personalized wrong question book through wrong question collection, automatic association of knowledge points, and intelligent analysis of wrong causes, to support students' targeted review.

5. The multifunctional integrated self-service printing learning machine according to any one of claims 1 to 4, characterized in that The learning machine includes a career planning intelligent prediction module, which predicts the career type suitable for students through interest and advantage subject analysis and decision tree model training, and provides personalized career planning suggestions; the learning machine also includes a mental health assessment and intervention module, which provides students with mental health assistance services through personalized psychological assessment, intelligent analysis of assessment results, and mental health counseling intervention.

6. The multifunctional integrated self-service printing learning machine according to any one of claims 1 to 5, characterized in that The learning machine uses NLP (natural language processing), large language model (LLaMA), RNN and similarity analysis algorithm to analyze the language attributes of the answers to Chinese and English language subjects, and compares the similarity with the standard answers; it evaluates the accuracy of answers to mathematics, physics, and chemistry science subjects through similarity analysis and clustering algorithms to ensure the accuracy and standardization of answer feedback.

7. The multifunctional integrated self-service printing learning machine according to any one of claims 1 to 6, characterized in that ,The learning machine utilizes a graph neural network (GNN) and a classifier (AdaBoost) to convert the ,students’ wrong questions and knowledge point data into a personalized ,knowledge graph, and generates a learning report that includes the ,students’ strengths and weaknesses analysis to assist in improving personalized ,learning.

8. The multifunctional integrated self-service printing learning machine according to any one of claims 1 to 7, characterized in that ,The learning machine includes a photo search and analysis module, which provides students with convenient question search and ,answering assistance by taking photos to identify question text, question matching and ,answering process association display technology.

9. The multifunctional integrated self-service printing learning machine according to any one of claims 1 to 8, characterized in that ,Based on the subject knowledge graph, the learning machine realizes personalized ,update of the knowledge graph, reasoning and prediction of the knowledge graph, and ,tracing application services of weak knowledge points.

10. An intelligent learning method based on the multifunctional integrated self-service printing learning machine according to any one of claims 1 to 9, characterized in that , including the following steps: S1. Construct a subject knowledge graph covering knowledge points of each subject, and establish a mapping between knowledge points and questions through automatic association and manual annotation; S2. Collect students’ answer data, use the improved AILBD algorithm to perform personalized learning diagnosis and analysis, and push personalized learning resources; S3, automatically collect students' wrong question data, and generate a personalized wrong question book through intelligent analysis of wrong question-related knowledge points and wrong reasons; S4. Based on students’ subject strengths and interest characteristics, use the decision tree model to predict their suitable career types and provide planning suggestions; S5. Conduct personalized psychological assessments for students and provide targeted psychological health counseling and intervention based on the analysis results; S6. Use natural language processing and knowledge graph technology to provide students with photo search and analysis services; S7. Dynamically update the personalized knowledge graph based on students’ answer feedback, and apply it to predict weak knowledge points and fill in the gaps.

Images