Parental controllable AI homework tutoring method based on E-book terminal

Abstract

The invention discloses a parent controllable AI homework tutoring method based on an E-book terminal. The method comprises the following steps: terminal binding and parent identity authentication; homework question collection and context analysis; calling a large model and scheduling an explanation strategy; mode control and authority path management; and the expression precipitation is shared with parent teaching plans. According to the method, a student terminal and a parent terminal are bound by using an eye protection display characteristic and a hardware controllability characteristic of an electronic paper book, and a learning behavior is ensured to be carried out in a visible and controllable range of parents through modes of code scanning authorization, equipment ID matching and the like; when the student requests to directly check answers or perform cross-border questioning, the system triggers parents to approve or block a path, and records the path to a behavior chain; in the normal learning process, the generative AI is called to generate multi-round and multi-style explanation content, and parents can select and settle the explanation content as a teaching plan for sharing. The homework AI tutoring system solves the problems of visual impairment, difficulty in supervision and risk of cross-border use in homework tutoring of the existing universal terminals such as mobile phones and tablet computers, and realizes an eye-protecting, safe and controllable homework AI tutoring mechanism.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence and education technology, in particular to a parent-controllable AI homework tutoring method based on an e-paper book terminal. The method utilizes the eye-protecting display characteristics and function-controllable characteristics of the e-paper book, combines the multi-round expression and personalized explanation capabilities of generative artificial intelligence, and constructs a controlled interaction mechanism suitable for primary and secondary school students' homework tutoring, so that students can use AI for learning within the scope of parental visibility and controllable permissions, avoiding the risks of eye damage and out-of-bound use caused by the use of general smart terminals such as mobile phones and tablets. BACKGROUND

[0002] In recent years, generative artificial intelligence technology (large models) has rapidly developed in the field of education, and students can directly ask AI questions and obtain homework analysis and problem-solving processes through natural language interaction. However, existing AI homework tutoring solutions mainly run on general electronic terminals such as smartphones, tablets, personal computers, etc. Although these devices have multimedia processing capabilities, they have the following common problems in the context of use by minors:

[0003] 1. Significant visual health risks Mobile phones and tablets mostly use backlit liquid crystal or OLED screens, which can cause visual fatigue, decreased vision, and other problems after long-term use. Parents, teachers, and education authorities generally oppose minors using such backlit screen electronic products for a long time.

[0004] 2. Difficulty in controlling learning boundaries General terminals can install various applications, and in addition to learning, they can also run social software, short videos, online games, shopping applications, etc., which makes it easy for students to be distracted when using the device for learning, and even access content unrelated to learning without the knowledge of parents.

[0005] 3. Lack of supervision of AI calls On existing general terminals, students can bypass parents and directly call large model services to obtain answers, lacking supervision and boundary control of the learning process, which can lead to dependent learning and affect the cultivation of thinking ability.

[0006] In comparison, e-paper books (electronic ink screen devices) are a type of eye-protecting reading terminal, which are more suitable for long-term focused reading and learning due to their characteristics of no backlight, low refresh, and reflective display. The display principle reduces the risk of visual fatigue, and naturally lacks multimedia entertainment functions such as video playback and high-frame-rate games at the hardware and system levels, making it widely recognized by parents, teachers, and society as a suitable learning terminal for students.

[0007] However, the existing e-paper book products are mainly used for electronic book reading in actual application, have high function openness, students can freely download and read materials, and lack content authority management and learning behavior supervision functions. Although this open mode limits the entertainment function, it still has problems such as uncontrollable reading content and lack of homework tutoring ability, so that the potential of e-paper books in the learning scene of primary and secondary schools is not fully realized.

[0008] Based on the above problems, the present application proposes a technical path of controlled e-paper book + large model AI homework tutoring:

[0009] The path does not limit the shape, brand of the e-paper book, nor the type of large model used, but creatively binds the core mechanisms of parent-controllable AI calling authority, homework collection, multi-modal explanation, cross-border content interception, behavior chain evidence, and the like, with the eye-protecting ink screen e-paper book hardware.

[0010] By realizing that the large model service can be called only under the authorization of parents in the e-paper book terminal, and limiting the installation and running of non-learning applications at the hardware and system level, students are completely prevented from accessing social, short video, game, shopping and other irrelevant content during the learning process.

[0011] As a “natural learning product” for students, the e-paper book can retain the eye-protecting advantage and become a safe entrance for AI-assisted learning in this controlled mode, thereby solving the dual pain points of uncontrollable general terminals and single function of traditional e-paper books. SUMMARY

[0012] I. Invention purpose

[0013] The purpose of the present application is to address the pain points of terminal bypassing supervision, lack of parental control in AI calling process, and untraceable learning behavior in existing student homework tutoring, and to propose a parent-controllable AI homework tutoring method based on an e-paper book terminal, which deeply integrates the closed and controllable characteristics of e-paper books with AI homework tutoring mechanisms to build a safe, focused, and supervised new family learning mode.

[0014] The present application binds the starting conditions, interaction process, permission intervention, and behavior record of AI homework tutoring on the e-paper book hardware, and combines the authorization mechanism of the parent terminal and cross-terminal real-time synchronization technology to achieve the following objectives:

[0015] 1. Guarantee the uniqueness and safety of the learning terminal Based on the closed system architecture and limited application ecosystem of e-paper books, students are prevented from downloading games and installing entertainment applications on the terminal, eliminating interference in the learning process from the hardware level. And under the authorization of parents, the student terminal e-paper book can call the AI homework tutoring function to prevent bypassing parents to directly obtain answers.

[0016] 2. Meet the multi-source homework collection needs Supports direct photo collection of homework on the student side of the e-paper book, and can also receive instant synchronization of photo topics from the parent side, ensuring that students complete learning on eye-protecting terminals throughout the process without switching to mobile phones, tablets, and other devices that are easy to distract.

[0017] 3. Optimize the eye-protecting presentation and interactive experience of AI explanation In combination with the low refresh mode of the e-paper book ink screen, the AI explanation content is optimized in terms of layout and contrast, and interactive methods such as annotation, underlining, and reading are supported, taking into account eye health and learning effectiveness.

[0018] 4. Strengthen real-time intervention and safety boundary control of parents Introducing two core control points, "answer disclosure right" and "boundary stop right", on the e-paper book side, parents can perform instant approval, content release, or interactive suspension through terminal binding, thereby accurately controlling the boundaries and depth of student learning.

[0019] 5. Realize the behavior chain record with traceability and tamper resistance Bind all interactive events to the e-paper book's unique hardware ID and family account to form a local and cloud dual evidence mechanism, achieving auditability and tamper resistance in the learning process, and providing a reliable basis for parental supervision, education management, and subsequent analysis.

[0020] Through the above technical solutions, the present application not only continues the inherent advantages of e-paper books in terms of eye protection and concentration, but also first uses the "closed and controllable" characteristics for the construction of safety boundaries in AI homework tutoring, forming a safe, controllable, and traceable AI homework tutoring closed loop in the family scenario, filling the technical gap of existing e-paper books in the field of intelligent homework tutoring, and having significant innovation and commercialization value.

[0021] II. Method step explanation

[0022] In the current primary and secondary education scenario, students' homework tutoring in the family environment generally faces two types of practical problems: on the one hand, parents have insufficient knowledge reserves and limited expression abilities in homework explanation and answer, making it difficult to provide continuous and effective tutoring; on the other hand, the rapid popularization of generative artificial intelligence enables students to directly obtain homework answers from AI, although it improves efficiency, it also brings learning dependency risks such as "directly copying answers without thinking", weakening the effect of independent problem solving and thinking training.

[0023] At the same time, more and more families tend to use e-paper books with eye protection ink screens as the only learning terminal for students for the sake of vision protection and concentration management. However, the existing e-paper book products, although having functions such as reading, note-taking, and photographing, lack a parent-controllable mechanism combined with AI homework tutoring, and cannot realize effective management of AI calling permission and real-time supervision of the learning process, resulting in a lack of visibility and intervention means for parents to children's learning behavior on such devices.

[0024] Based on this, the present application proposes a parent-controllable AI homework tutoring method based on an e-paper book terminal, which binds the calling, presentation, intervention and behavior recording of AI homework explanation on the e-paper book hardware, combines the parent authorization mechanism and cross-terminal synchronization technology, meets the needs of students to learn on eye protection terminals with high quality, and ensures the control of parents on learning boundaries and key nodes. The method takes e-paper book as the only learning terminal for students, and through multi-source homework collection, AI eye protection rendering, real-time intervention of parents and behavior chain notarization, a safe, controllable and traceable homework tutoring closed-loop system is constructed.

[0025] Step one: device binding and parent authorization login

[0026] This step is used to realize the safe binding of the student end e-paper book and the family account and the authorized login control of the parent end, and to ensure that the AI homework tutoring function cannot be started by the student alone in the unauthorized state from the hardware level.

[0027] Specifically, the system first reads the device unique identification information in the student end e-paper book, including but not limited to device serial number, MAC address, hardware fingerprint code or device ID generated by a security chip, and establishes a binding relationship between the identification information and the family account. The binding process can be realized through the registration operation when connecting for the first time, and the binding request is initiated by the parent end and the confirmation is completed.

[0028] After the binding relationship is established, the student end e-paper book needs to be authorized by the parent end device to enter the available state. The parent end device can be a mobile terminal (such as a smart phone, a tablet computer), or another parent end e-paper book that has been bound. The authorization methods include but are not limited to two-dimensional code scanning authentication, face recognition authentication, fingerprint recognition authentication, password input authentication, etc., and the system can select a single or multi-factor combination verification mode according to the security requirements.

[0029] After authorization is successful, the system will generate a one-time authorization token (Token) or session key and issue it to the student end e-paper book. The student end is only allowed to call the service interface related to AI homework tutoring within the validity period of the token; when the token is invalid or revoked, the student end AI tutoring function immediately enters the disabled state, and the interface displays a prompt message that re-authorization is required.

[0030] This step ensures that students cannot independently use the AI homework tutoring function without the permission of the parents by binding the physical hardware ID of the student's e-paper book with the family account and introducing a login authorization process for the parent's end, thereby achieving security control and usage permission management based on hardware binding.

[0031] Step two: homework collection and multi-source input

[0032] This step is used to achieve multi-source collection and input processing of homework problems between the student's e-paper book and the parent's device, ensuring that students can obtain complete homework content in an eye-protecting terminal environment and supporting multiple data input methods to adapt to different usage scenarios.

[0033] Specifically, the student's e-paper book preferably has a built-in camera collection module for directly capturing paper homework problems or test paper content. The image data after shooting is pre-processed locally, including image distortion correction, brightness and contrast adjustment, and noise filtering, to ensure the accuracy of subsequent recognition.

[0034] At the same time, the system supports the parent's device as an auxiliary collection entry, and the parent can use a mobile phone, tablet, or parent's e-paper book to take a photo to obtain homework problems and transmit image data to the student's e-paper book through the synchronization channel bound by the family account. The student's e-paper book automatically pops up a homework preview interface when receiving the synchronization data, allowing the student to confirm and enter the subsequent AI explanation process.

[0035] In addition to camera collection, the student's e-paper book also integrates electromagnetic handwriting collection and touch input functions, allowing students to directly write and input problem content on the eye-protecting screen or input text through a soft keyboard. Handwriting input data is locally vectorized and structured, and text input is directly transmitted to the analysis engine.

[0036] Regardless of the collection approach, the system locally calls the optical character recognition (OCR) module and / or handwriting recognition module on the student's e-paper book to convert homework images or handwriting into structured problem data, and the recognition results include problem number, problem type, text content, and optional image / formula region annotation information.

[0037] Through the above multi-source input design, this step ensures that students can rely on the e-paper book hardware to complete homework content acquisition without using other non-eye-protecting terminals, and supports cross-end collaboration between the parent's end and the student's end in the collection stage, providing a complete problem input basis for subsequent AI explanation and permission control.

[0038] Step three: AI explanation and eye-protecting rendering

[0039] This step is used to generate AI model based on student end e-paper book call homework explanation content in the state of parental authorization, and optimize rendering in e-paper book eye protection display environment, to meet the visual health and interaction needs in the process of long time learning.

[0040] After completing the homework collection and structured analysis of step two, the student end e-paper book will package the parsed question information into a standardized call request, which includes question text, question type label, subject information, grade or difficulty level, and user selected explanation style parameters. The call request is only allowed to be sent to the AI explanation server under the condition that the valid authorization token generated in step one exists and passes the verification.

[0041] After receiving the request, the AI explanation server calls one or more artificial intelligence reasoning models matched with the question type and subject based on the preset strategy template and multi-round generation mechanism, generates multi-step explanation content including problem solving ideas, detailed steps, key knowledge point prompts and precautions. The generation result is returned to the student end e-paper book in a structured data format.

[0042] After receiving the explanation result, the student end starts the eye protection rendering process. This process first optimizes the layout of the explanation content on the low refresh rate ink screen, including but not limited to: text line spacing and font weight adjustment, key step highlight gray scale marking, formula and graphics gray scale anti-aliasing processing, etc. During the rendering process, the system also supports the annotation superposition function, allowing students to use the electromagnetic pen to underline, circle or add annotation text directly on the explanation content, and the annotation data is saved synchronously with the original explanation content.

[0043] In addition, the student end can select the reading mode, which converts the explanation content into voice by the built-in speech synthesis engine, and plays it through the e-paper book speaker or external earphone, realizing the multi-modal learning experience of listening and learning. The reading mode supports step-by-step playback and re-reading function, which is convenient for students to repeatedly listen to the relevant explanation when they master the difficult points.

[0044] Through the design of this step, the explanation content generated by AI not only technically binds with the e-paper book hardware, ensuring the call under the controllable condition of parents, but also fully utilizes the display characteristics of the eye protection ink screen in the presentation form, providing a visual comfortable and natural interactive learning environment for students.

[0045] Step four: parental control and permission intervention

[0046] This step is used to introduce the parental permission control mechanism in the process of AI homework tutoring on the student end e-paper book, through the two core intervention points of "answer revealing right" and "boundary stopping right", to realize the active supervision and immediate intervention of the student learning process, and ensure the compliance of the tutoring behavior and the controllability of the learning goal.

[0047] Specifically, when the student end e-paper book requests to directly view the answer, the system will immediately pause the current AI interaction process and generate an answer disclosure approval request data packet containing the question identification, request time, question type information, and current session identification. The data packet is sent to the parent end device (including the parent's mobile phone or the parent end e-paper book) through the synchronization channel bound to the family account. After receiving the approval request, the parent end interface will display the question preview and optional approval options, including "allow to display the answer" and "reject to display the answer and continue explanation only". The parent's approval instruction is returned to the student end e-paper book after signature verification, and the system executes the corresponding content presentation or continues the explanation process according to the instruction.

[0048] When the content input or photographed by the student end e-paper book is determined by the system to be non-learning information (such as entertainment, games, or pictures or texts unrelated to homework) or is recognized as an overstepping behavior that bypasses the established prompt to directly call AI, the boundary detection module will immediately trigger the boundary stopping mechanism. This mechanism includes freezing the current AI service user interface of the student end e-paper book, blocking the calling link with the AI service server, and displaying prompt information on the screen waiting for the parent to handle. At the same time, the system generates a boundary event record containing boundary type, trigger source, timestamp, terminal hardware ID, etc. and sends it to the parent end device through the synchronization channel.

[0049] After receiving the boundary notification, the parent end can view the event details in the interface and execute handling instructions, including "unfreeze and continue tutoring", "terminate the current session", or "delay unlocking". All parent handling results will be synchronized to the student end e-paper book and written into the behavior chain record bound to the hardware ID for subsequent auditing and backtracking.

[0050] Through this step, parents can intervene in the student's learning process in a remote or local manner, making AI homework tutoring no longer a one-way autonomous behavior of the student, but a family tutoring collaboration mode running within a controllable boundary. At the same time, this permission intervention process is bound to the e-paper book hardware identity, ensuring that only authorized parent end devices can control the student end learning behavior, effectively preventing unauthorized intervention or bypassing the parent's violation of use.

[0051] Step five: Real-time synchronization and behavior chain record

[0052] This step is used to realize the real-time synchronization of the screen and interaction process between the student end e-paper book and the parent end device during AI homework tutoring, as well as the generation and storage of the whole process behavior chain, so as to ensure that parents can intuitively understand the student's learning dynamics at any location, and ensure that the learning data has traceability and tamper resistance.

[0053] Specifically, after starting the AI tutoring session, the student-side e-paper book synchronizes the screen rendering frames and user interaction events (such as page turning, annotations, answer submission, etc.) to the parent-side device in real time through a secure and encrypted home account synchronization channel. The parent-side can display the current screen of the student-side in mirror mode on the local interface, achieving a millisecond-level delay "what you see is what you get" remote supervision effect.

[0054] To ensure the integrity of the behavior record, the system locally establishes a behavior chain record structure on the student-side e-paper book. This structure stores node data including the following fields in time series: timestamp, terminal hardware unique ID, home account identifier, session identifier, operation type (photographing, annotation, answer request, etc.), event parameters (question identifier, explanation mode, annotation content summary, etc.), parent approval instruction and execution result, etc. All node data are correlated through chain hash pointers to prevent insertion or deletion in the middle.

[0055] At the same time of generating the behavior chain record, the system encrypts and signs the node data and periodically uploads the local behavior chain snapshot to the cloud storage, forming a dual evidence mechanism of local and cloud. Once the local record and the cloud snapshot are inconsistent, the system can immediately trigger an abnormal alarm and mark the session as a high-risk session.

[0056] Any control instruction executed on the parent-side (including answer release, boundary crossing stop, session termination, etc.) is also written as an independent node to the behavior chain, ensuring that all intervention behaviors have clear responsibility attribution and time record. Through this design, even if the student changes the account or resets the device, as long as the behavior occurs on the same e-paper book hardware, its unique hardware ID will be recorded in the behavior chain, making it impossible to evade supervision and audit.

[0057] This step realizes the full-process visualization and data traceability of the student's learning process, not only providing real-time supervision means for parents, but also providing a reliable data foundation for subsequent learning behavior analysis, dispute evidence collection and compliance audit.

[0058] III. Implementation process overview

[0059] The present application provides a parent-controllable AI homework tutoring method based on a controlled e-paper book terminal, which solves the visual health risks, content boundary risks and learning supervision deficiencies of general terminals such as smartphones and tablets in student learning scenarios by deeply binding AI calls with eye-care e-paper book hardware. This method does not limit the brand or shape of the e-paper book, nor does it rely on specific large model types, but creatively builds a safe, controllable and traceable AI learning path, meeting the common demands of parents, teachers and society for healthy use of devices and control of learning boundaries for minors.

[0060] First, device binding and parental authorization login. The student e-paper book is bound to the family account by reading the unique hardware identifier at the factory or first use, forming an irreplaceable device identity. The parent end (which can be a mobile phone or another parent e-paper book) authorizes login through safe methods such as scanning codes, facial recognition, or passwords, and the generated authorization token allows the student end to call the AI tutoring function within the valid period. This design ensures that students cannot use AI services without the permission of parents, forming a controlled entrance for e-paper books.

[0061] Second, assignment collection and multi-source input. The student e-paper book can directly take photos of assignments through the built-in camera module, or input questions through electromagnetic handwriting or text input, while supporting parents to take photos and synchronize them to the student end in real time. This multi-source collection capability allows students to complete assignment input on eye-protecting devices throughout the process, avoiding the use of mobile phones and tablets that pose a risk of boundary crossing, and meeting the core needs of parents who want their children to learn only on safe terminals.

[0062] Third, AI explanation and eye protection rendering. Under the authorization state, the student e-paper book can call any compatible large model to generate multi-step explanation content, including problem-solving ideas, steps, and knowledge point prompts. The explanation content is displayed after being optimized for low refresh on the ink screen, and supports interactive operations such as annotation, underlining, and reading aloud. This not only takes full advantage of the eye-protecting features of e-paper books, but also ensures that the presentation and interaction of AI-generated content are carried out in a controlled environment, eliminating the interference of entertainment functions and irrelevant information.

[0063] Next, parental control and permission intervention. During the learning process, parents can approve the student's direct answer request through the answer reveal right, or block the call of non-compliant content through the boundary stop right. Once the boundary detection is triggered, the student e-paper book interface is immediately frozen, waiting for the parent's instruction. This dual-control point mechanism is the core advantage of controlled e-paper books over existing open terminals, responding to the concerns of society and the education sector about the safety of minors using self-determined devices.

[0064] Finally, real-time synchronization and behavior chain recording. The screen image and operation events of the student e-paper book are mirrored to the parent end in real time through an encrypted channel, realizing "what you see is what you get" remote supervision. At the same time, all interactive events are bound to the unique hardware ID of the e-paper book and the family account, forming a chain of behavior records, and storing evidence locally and in the cloud. This ensures the traceability and tamper resistance of the learning process, and provides a reliable basis for learning behavior for parents, teachers, or schools.

[0065] Through the above steps, the controlled electronic paper book is promoted from a single reading tool to a learning terminal with parental supervision, boundary protection, and AI efficient tutoring, realizing controllable AI homework tutoring in an eye-protecting and safe hardware environment, which meets the health requirements of students using the machine for a long time, and also meets the strict demands of parents and society for the safety and learning boundaries of the learning terminal.

[0066] IV. Model calling structure and boundary protection mechanism

[0067] In the AI homework tutoring process of the electronic paper book terminal, a multi-link and multi-role collaborative model calling and boundary protection structure is established to ensure that the learning process of students is within the controllable range of parents and prevent behaviors such as directly copying answers, crossing boundaries to ask questions, or evading supervision. This mechanism relies on the natural controlled hardware properties of the electronic paper book, forms a collaborative closed loop with the parent terminal, the binding relationship of the family account, and the multi-model scheduling module, and specifically includes the following three aspects:

[0068] 1. Calling path structure

[0069] In the present application, the AI calling flow of the electronic paper book terminal strictly follows the following path:

[0070] Step one: Call initiation The student initiates a homework tutoring request on the electronic paper book terminal, which can include multiple sources such as shooting questions, handwriting annotations, and voice questions.

[0071] Step two: Permission verification The call request first enters the permission verification module to verify whether the terminal is a student terminal under the binding family account and check the login status of the parent.

[0072] Step three: Behavior compliance check The system performs boundary detection on the input content, including identifying whether it is a direct request for answers, non-learning text or images, and sensitive content, etc.

[0073] Step four: Strategy template matching For compliant requests, the system selects a prompt template and expression strategy based on the question type, subject, and parent-set explanation preferences; for boundary-crossing requests, it enters the parent approval or prompt reconstruction process.

[0074] Step five: Model calling and generation Call the adapted large model (including but not limited to mathematics, Chinese, English, etc. subject models) to generate explanation content, ensuring that the output meets the learning purpose.

[0075] Step six: Eye protection rendering and output The generated content is rendered into a low-refresh, high-contrast eye protection mode suitable for e-paper book display and displayed on the student terminal, while being synchronized in real time to the parent terminal.

[0076] 2. Boundary protection mechanism To prevent abuse and illegal use, the present application introduces two core protection mechanisms on the model call chain: Answer disclosure approval mechanism: when the system detects that the student directly requests an answer, the parent's approval is triggered immediately, and the parent can choose "confirm the answer" or "guide the topic" on the terminal, and the approval result takes effect immediately and is recorded in the behavior chain; Boundary detection and blocking mechanism: non-learning inputs such as game guides, entertainment content, and irrelevant pictures are blocked, and warning information is pushed to the parent terminal, while providing compliant prompts or restructured learning requests for students to avoid interference with the learning process.

[0077] 3. Anti-circumvention design The present application ensures that the system cannot be simply replaced or circumvented through hardware binding and call chain closed design: Hardware binding: each model call needs to verify the binding relationship between the unique device ID of the e-paper book and the family account, and the terminal without binding cannot directly call the AI tutoring function; Call chain closed loop: regardless of the type of large model called (including third-party or self-developed models), it needs to go through the permission verification and strategy template link to ensure that illegal calls that bypass the check cannot generate valid output; Multi-model adaptation: the system supports dynamic switching of models by subject and type, but does not change the call structure and permission path to prevent circumventing patent protection by replacing model providers.

[0078] Through the above structural design, the present application not only utilizes the hardware characteristics of e-paper books to achieve a "naturally controlled" AI homework tutoring environment, but also establishes a model call and boundary protection mechanism that is auditable, traceable, and blockable at the software level, ensuring the safety and compliance of the student learning process, and fully meeting the needs of parents for learning supervision and content boundary control.

[0079] Five, term definition and identification boundary explanation

[0080] To avoid semantic ambiguity during patent implementation and examination, and to ensure that the protection scope of the present application is clear and has expandability, the key terms appearing in the specification and claims are now defined. Unless otherwise stated, the terms described in this section have the specific meaning of the present application and do not limit their conventional use in other technical literature.

[0081] 1. E-paper book terminal

[0082] Smart learning terminal with eye-care display characteristics (such as electronic ink screen) and supporting controlled interaction, including but not limited to e-book readers, electronic ink tablets, customized learning devices integrated with eye-care display screens, etc.

[0083] In the present application, the electronic paper book terminal has a function-limited characteristic, i.e., it does not have the ability to install arbitrary third-party applications, and cannot be used for non-learning purposes (such as video playback, games, shopping, entertainment socializing, etc.), and is ensured to run only learning applications authorized by parents or the system through system-level permission control.

[0084] The electronic paper book terminal can be built-in or externally connected with functions such as photographing collection, note annotation, and voice interaction, but the invocation of the above functions is subject to the constraints of the parental permission control mechanism.

[0085] 2, Parental terminal

[0086] Refers to a smart terminal device used to manage and control student learning behavior, including but not limited to smartphones, tablets, computers, electronic paper books, or other devices that can run management applications.

[0087] The parental terminal is the master control device of the behavior chain in the present application, and has core control permissions such as binding, authorization, monitoring, approval, and suspension of student terminals, and all student terminal behaviors are recorded under the parental terminal identity chain.

[0088] 3, Student terminal

[0089] Refers to a terminal device bound to the parental terminal for students to perform homework learning and AI interaction, including but not limited to electronic paper books, tablets, computers, and special function learning machines, etc.

[0090] The student terminal does not have independent permissions in the present application and must be authorized by the parental terminal before it can be logged in and used, and its functions are limited by the system and can only perform learning-related operations.

[0091] 4, Answer disclosure right

[0092] Refers to when a student requests to directly view the answers to homework, the system needs to initiate an authorization request to the parental terminal, and the parent decides whether to directly display the answers or replace them with AI-generated explanations.

[0093] The answer disclosure right is one of the core permission control points of the present application, ensuring that students use AI learning under the controllable conditions of parents and preventing reliance on direct copying of answers.

[0094] 5, Suspension right

[0095] Refers to the right of parents to interrupt the AI interaction or homework tutoring process at any time during the use of the student terminal, especially when the student asks questions or operates beyond the boundaries (such as uploading non-learning content or requesting non-learning questions), the parent can terminate the behavior chain with one click.

[0096] 6. Boundary-crossing questions

[0097] Refers to students asking AI questions unrelated to homework learning or violating parental / system preset boundaries, including but not limited to questions in the fields of entertainment, games, shopping, socializing, movies, and requests that may involve privacy or sensitive content.

[0098] The system can automatically identify boundary-crossing questions and trigger a blocking and parental approval mechanism.

[0099] 7. Prompt template

[0100] Refers to the system's preset sentence structure or expression paradigm that can guide students to ask AI questions in a compliant manner, such as "Please help me analyze the problem-solving approach for this question" or "Please explain this passage in a life-like way."

[0101] Prompt templates can be provided by the system or customized by parents, and can be dynamically adjusted according to grade, subject, and learning goals.

[0102] 8. Model invocation chain

[0103] Refers to the ordered chain of steps involved in the entire process from the student / parent issuing a question or task to the AI generating content and returning, including model selection, strategy scheduling, prompt processing, style generation, etc.

[0104] In this invention, the model invocation chain can support multiple model collaboration at the same time, and perform permission judgment and boundary check during the invocation process.

[0105] 9. Behavior chain record

[0106] Refers to the process of the system storing each valid behavior (such as asking questions, approving, stopping, generating explanations, etc.) of the student terminal and the parent terminal in the learning process in a chain.

[0107] The behavior chain contains timestamp, terminal ID, initiator identity, prompt content, invoked model, output type, approval result, etc. core fields, and is always attributed to the parent terminal identity, forming a traceable and auditable record system.

[0108] 10. Scope of protection and non-involved field explanation

[0109] To clarify the technical positioning and protection boundaries of this invention and avoid semantic misunderstandings during the review and implementation process, the following is declared:

[0110] 1) Non-involved fields The technical solutions described in this invention: Do not involve designing, building, training, or optimizing any artificial intelligence (AI) model itself; No adjustment, fine-tuning or updating behavior of internal parameters of AI model is involved; The algorithm implementation, corpus formation or generation mechanism of the AI model is not claimed to be protected; No control behavior is constituted on the operation principle or internal structure of the AI model; No originality judgment of education content itself, knowledge correctness verification or test question answer generation method is involved.

[0111] 2) Protection core The protection focus of the present application is: AI tutoring call structure design based on controlled e-paper book and other controlled hardware terminals; Parent-led permission control and approval mechanism, including answer disclosure right and boundary-crossing stop right; Strategy path management and boundary constraint before and after AI call; Behavior chain and version tree management mechanism of learning process, realizing traceability, auditability and sedimentation; Structured implementation of parent teaching plan sharing, referencing and incentive mechanism; The type, manufacturer, brand or implementation of the called AI model is not limited, and the hardware shape and brand are not limited.

[0112] 3) Technical focus The present application focuses on using controlled e-paper book and other low-interference, eye-protecting and supervisable special terminals to solve the problems of supervision difficulty, high boundary-crossing risk and insufficient learning concentration of traditional mobile phones, tablets and other devices in the AI tutoring scene of homework. The terminal and the parent control mechanism form a closed loop, so that students can efficiently use AI for learning within a safe and controllable boundary.

[0113] Six, path control and abnormal bypass blocking mechanism

[0114] In the process of realizing parent-controllable AI homework tutoring based on e-paper book terminal, in order to prevent students from generating answers by avoiding prompt language, skipping parent approval process or directly calling AI model, the path control and abnormal bypass blocking mechanism is specially designed. This mechanism combines the parent authorization chain, behavior chain record and the controlled environment unique to e-paper book terminal to realize the whole-process control of model call path, ensuring that the tutoring process runs within the compliance boundary of auditability and traceability.

[0115] 1, path control structure The system includes all AI call behaviors on the e-paper book terminal into a unified path control structure, and each homework explanation request needs to go through the following steps: Prompt language binding verification: after receiving the homework question, the system first judges whether it is accompanied by a compliance prompt language template; if not, it automatically blocks the call and guides the student to select or generate a prompt language.

[0116] Permission node verification: Nodes involved in the call path that "directly reveal the answer" must trigger the parental approval interface, allowing parents to choose "allow display of answers" or "change to explanation mode."

[0117] Boundary crossing type identification: The system analyzes input text and image content in real time. When it detects irrelevant boundary crossing questions (such as game guides, entertainment information, etc.) unrelated to homework, it directly enters the blocking process.

[0118] 2. Abnormal bypass detection The system has the ability to detect the following abnormal behaviors: Skip prompt call: Students directly input questions without triggering the strategy template; Cross-terminal bypass: Students attempt to initiate AI calls on non-bound terminals; Multi-round evasion generation: After being rejected, students guide AI to output answers through multiple rounds of dialogue; Image boundary crossing request: Upload screenshots and photos unrelated to homework (such as game screenshots, social chat records, etc.).

[0119] 3. Blocking and approval mechanism Blocking action: When abnormal bypass behavior is detected, the system immediately blocks the current call request and displays a blocking prompt on the student terminal.

[0120] Parental approval wake-up: The system synchronously pushes an approval request to the parent terminal, including blocking reasons, original request content, and risk level; parents can choose to release, rewrite the prompt and release, or maintain the block.

[0121] Path recovery: After the parents release, the system restores the call path and records the approval action in the behavior chain; if refused, the call process terminates.

[0122] 4. Behavior chain and blocking record All blocking events and approval results are written into the behavior chain owned by the parents, including: Abnormal type (skip prompt, cross-terminal, boundary crossing question, etc.); Abnormal detection time and terminal ID; Approval person identity and approval result; Call path hash digest for subsequent auditing and tracing.

[0123] 5. Combination with e-paper hardware features The electric paper book terminal limits application installation and network access range at the system level, and only allows interaction with the AI model through the tutoring application authorized by the application, and the path control mechanism is combined with the hardware security strategy to prevent students from bypassing the system call through a browser, a third-party APP or an offline file, and to block the abnormal bypass risk from the source.

[0124] Through the path control and abnormal bypass blocking mechanism, the application not only realizes technical protection of the student AI use process, but also realizes the "answer disclosure right" and "stop right" of parents into an executable and traceable control link, so that safe, compliant and supervised AI homework tutoring is realized in the controlled environment of the electric paper book.

[0125] Seven, beneficial effects

[0126] The application realizes the whole-process boundary supervision of the student learning process and the controllable management of parents by introducing a closed controllable architecture based on the electric paper book terminal in the homework tutoring scene, and has the following significant beneficial effects:

[0127] 1. Natural closed running environment, reducing the risk of crossing the border Compared with open terminals such as mobile phones, tablets and computers, the student terminal of the electric paper book is based on an electronic ink screen and a limited operating system, which physically shields non-learning functions such as video playback, games and social networking. Students cannot install applications or bypass external services, which greatly reduces the risk of indulging in entertainment, information crossing and learning distraction.

[0128] 2. Hardware-level binding and whole-process authorization control of the parent terminal The student terminal of the electric paper book must complete hardware-level binding and identity authorization through the parent terminal, and must pass dynamic verification each time it logs in. Key behaviors during the learning process (such as directly requesting answers and uploading non-learning content) must trigger parent approval to ensure that parents have absolute "release rights" and "stop rights" for student AI interaction.

[0129] 3. Real-time mirror supervision and behavior chain traceability The parent terminal can mirror the learning interface and AI interaction process of the student terminal in real time, and all events such as questioning, approval and blocking are recorded as a behavior chain with the parent as the responsible subject, realizing auditable and traceable whole-link supervision to meet the requirements of family supervision and education compliance.

[0130] 4. Controllable output of AI explanation and teaching plan sedimentation All AI-generated explanation content must be authorized by the parent terminal before being displayed on the student terminal or broadcast through the auxiliary terminal, and can be annotated with subject, grade and question type tags by the parent one-key sedimentation as a "parent teaching plan" for subsequent review and cross-family mutual assistance and sharing, forming a sustainable and reusable high-quality education resource library.

[0131] 5. Commercialization and education scene are highly consistent Relying on the eye protection characteristics and closed controllable attributes of the electric paper book, the application can be landed in school, training institution, family and other scenes, which not only solves the pain points of parents "remote monitoring is difficult and children are easy to cross the border", but also meets the needs of educational institutions for "controllable AI learning terminal", has clear market promotion potential and differentiated competitive advantage. BRIEF DESCRIPTION OF DRAWINGS

[0132] Figure 1 : the overall structure of the system is shown in the figure; Figure 2 : the structure of the closed controllable learning environment of the electric paper book is shown in the figure; Figure 3 : the flow chart of family account binding and terminal role permission management; Figure 4 : the flow chart of classroom extension type tutoring scene; Figure 5 : the flow chart of local strategy guarantee under weak network environment; Figure 6 : the answer disclosure approval and cross-border stop mechanism diagram; Figure 7 : the behavior chain record and synchronization mechanism diagram. DETAILED DESCRIPTION

[0133] In order to make the purpose, technical scheme and beneficial effects of the application more clear and definite, the application will be further described in detail below in combination with the drawings and examples. The application is not limited to the following specific examples, any equivalent replacement or improvement within the spirit and principles of the application shall be included in the protection scope of the application.

[0134] Example one: parent tutoring type (mode one) homework explanation based on electric paper book

[0135] In this embodiment, the parent terminal is a smart phone and the student terminal is an electric paper book reader. The parent uses the mobile phone App to shoot the homework questions and upload them to the system, and the system performs OCR recognition and question type analysis on the cloud to structure the questions into context information including subject, grade, knowledge point and other labels. After receiving the explanation task, the electric paper book terminal displays the step-by-step explanation content of the generative AI through the electronic ink eye protection screen. Since the electric paper book is a closed and controlled system, it does not support third-party application download, and students cannot bypass the parent to install games or non-learning software in the terminal, so as to ensure that the use process is limited to homework tutoring. The parent can master the explanation content and progress throughout the process, and can deposit high-quality explanation as "parent teaching plan" when needed.

[0136] Example two: student self-directed type (mode two) AI learning based on electric paper book

[0137] In this embodiment, after the student is authorized by the parent by scanning the code, the student can independently use the AI tutoring function on the e-paper book end. The student can take a picture of the homework on the e-paper book, call AI to generate an explanation, and realize operations such as “change the way of saying” and “repeat again” through the built-in prompt function. When the student requests to directly view the answer, the system will push an approval request to the parent's mobile end, and the parent will decide whether to directly release or guide the AI to explain the question. The whole learning process is displayed in real time on the parent's end, and the parent can interrupt the boundary behavior at any time. Since the e-paper book system is closed, it cannot install entertainment, social, and game applications, and students cannot bypass the system to engage in non-learning activities when they are alone, ensuring the controllability and peace of mind of parents in the learning process.

[0138] Example Three: Mistake Review and Variant Training on E-Paper Book End

[0139] In this embodiment, the student takes a picture of the entire test paper on the e-paper book, and the system automatically identifies the question location, question type, and score, analyzes the correct rate, and generates a mistake distribution. For the mistakes, the system generates a mistake analysis on the e-paper book end and automatically pushes variant training questions corresponding to the knowledge points. The student directly answers on the e-paper book, and the system instantly grades and explains until mastery. The parent end can view the mistake training progress and results, and include the completed training package in the subsequent review plan. Since the e-paper book only supports learning-related applications and does not have a browser or application store, students will not be disturbed by games, videos, and other distractions during long-term review, making the learning process more efficient and controllable.

[0140] Example Four: E-Paper Book Learning Closed Loop under Parent Remote Supervision

[0141] In this embodiment, the student uses the e-paper book at home for homework tutoring, and the parent monitors the student's learning process in real time through the mobile end when they are out. The e-paper book has a built-in parent binding account, and all AI explanations, question taking, answering, and mistake training operations are mirrored in real time to the parent's mobile end. If the system detects that the student inputs text or pictures unrelated to the homework (such as game screenshots or entertainment topics), it will immediately block the request and push a “boundary behavior reminder” to the parent. The parent can stop the student's operation with one click, or choose to approve it to continue. Since the e-paper book hardware system is designed in a closed manner and does not have the function of independent network download and third-party application installation, even when the parent is not present, it can prevent students from bypassing to non-learning scenarios from the hardware level. This design not only solves the pain points of parent remote supervision, but also improves the market's acceptance and trust of AI homework tutoring hardware.

[0142] Example Five: Voice Interaction Tutoring for Low-Grade Students

[0143] In this embodiment, the electric paper book connects with the smart voice doll through Bluetooth, and is used for interesting homework guidance for students in lower grades. After the parents upload the questions, the electric paper book displays the explanation content, and at the same time allows the students to click the “voice play” button to broadcast the explanation content in a voice style that children like. The students cannot make the doll broadcast unauthorized content by themselves, because the electric paper book terminal will perform permission verification on all play requests, and only the explanation content with the “broadcastable tag” can be sent to the doll. This function not only retains the eye protection feature, but also improves the learning enthusiasm of young users through an interesting learning method, and forms a differentiated selling point in the family education scene. For hardware manufacturers, this scheme has strong commercialization and promotion potential, because the combination of the electric paper book and the smart doll can build a complete family education product ecosystem.

[0144] Embodiment six: Electric paper book synchronous push of school centralized arrangement of homework

[0145] In this embodiment, the school teachers can arrange homework in batches through the teacher terminal platform, and the system pushes the homework to the electric paper book terminal of the students. The students use the electric paper book to directly perform homework question uploading, AI explanation, error question training and other operations, and the completion is automatically fed back to the teacher and the parent terminal. This integrated mode can enable the school to participate in the homework guidance closed loop, forming a four-way cooperation of “teacher arrangement-student learning-parent supervision-AI guidance”. The electric paper book can be used as the only learning device in the school and the family, and there is no risk of entertainment application of mobile phones, tablets and other devices, thereby greatly reducing the concern of parents and teachers about the distraction of students during homework time. This mode has commercialization potential for cooperation with schools, and can realize income through education hardware procurement, family value-added services and the like.

[0146] Embodiment seven: Pay value-added service and family education subscription mode

[0147] In this embodiment, the electric paper book terminal not only provides basic AI homework guidance function, but also can enable high-level function modules such as error question variant training, special knowledge point breakthrough, learning situation analysis report, etc. through subscription service. The parents can purchase the corresponding package according to the learning needs of the children, and the system automatically unlocks the service content on the parent terminal and the electric paper book terminal. Due to the closed nature of the electric paper book system, students cannot crack or bypass the paid function through non-official channels, ensuring the stability of the business model and the protection of intellectual property rights. This design provides a sustainable commercial profit mode for manufacturers, and increases the long-term stickiness of family users.

[0148] Embodiment eight: Classroom extension type parent controllable AI homework guidance scene

[0149] In this embodiment, the system runs on the student e-paper book terminal and the parent mobile terminal bound to the family account. In class, the teacher pushes the homework content to the parent App through the homework assignment platform or teaching software. The parent terminal receives the homework task information in real time and synchronizes it to the student e-paper book after identity authentication confirmation.

[0150] After school, the student opens the homework task on the e-paper book, and the system automatically enables the AI homework tutoring function according to the parent's authorization status. All behaviors such as question viewing, prompt calling, and answer requesting during the tutoring process are executed within the e-paper book closed system, and there is no possibility of switching to entertainment applications or external browsers, eliminating the risk of learning distraction and bypassing the answer from the source. The parent terminal can mirror the student's learning process in real time, and when the student requests to directly view the answer, the parent can make a "release" or "guided explanation" approval decision, and the approval result takes effect immediately on the e-paper book. The entire behavior chain is bound to the e-paper book's unique hardware ID and family account, ensuring that it can be audited and tamper-proofed in the future.

[0151] Embodiment Nine: Local Strategy Guarantee Scene in Weak Network Environment

[0152] This embodiment is suitable for students in a family or boarding environment where Wi-Fi is unstable or the network is interrupted. The student e-paper book is pre-authorized by the parent e-paper book in a networked state for AI homework tutoring tasks, and the corresponding question data, prompt template, and explanation strategy package are encrypted and stored in the e-paper book's local controlled area.

[0153] When the student is in a network-free or weak network environment, the e-paper book can directly call the authorized AI explanation strategy locally to explain and interact with the specified homework questions step by step. All local tutoring behavior chains (including question calling, answer application, and parent pre-set approval decisions) will be encrypted and recorded in the e-paper book's local storage.

[0154] Once the network is restored, the e-paper book will automatically synchronize the behavior chain records to the parent terminal and the cloud platform, and the parent can completely trace the student's offline learning operations, ensuring that the closed tutoring mode of "parent controllable and fully traceable" is maintained even in unstable network conditions.

Claims

1. A parent-controllable AI homework tutoring method based on an electric paper book terminal, characterized in that, Comprise the following steps: Terminal binding and parent identity authentication: through the parent terminal, the student terminal of the electronic paper book is uniquely bound at the hardware level, and the binding identifier includes but is not limited to device ID, MAC address, fingerprint information or serial number, and the first authorization and dynamic verification each time are completed through parent identity authentication (face recognition, password, code scanning, etc.); The electronic paper book is a closed and controllable learning terminal, and the system level shields the installation and running of other applications except learning function; Collection and context analysis of homework questions: the student terminal or the parent terminal collects homework question information, including photographing, scanning or handwriting input, and analyzes the question type, subject label and context information; AI explanation strategy calling and permission control: the system calls one or more large models to generate explanation content within the permission boundary authorized by the parent, and when the student requests to directly view the answer or makes an out-of-bound question, the parent approval process is triggered, and the parent can choose to release, generate an explanation substitute or refuse; Path control and abnormal bypass lock: in the process of calling AI on the student terminal, it is detected whether there is an abnormal path such as skipping prompt, bypassing authorization, calling unbound models, etc., and if found, it is immediately locked and the behavior chain is recorded; Explanation content sedimentation and controllable output: the explanation process confirmed by the parent is stored as a parent teaching plan, and the student terminal can display or play through the auxiliary terminal in a controlled interface, and all outputs are subject to the permission limit of the parent terminal and the system level closed mechanism.

2. The method of claim 1, wherein, The terminal binding adopts one-time code scanning or input binding code to establish the pairing relationship between the parent terminal and the student terminal, and dynamic secondary verification is performed each time to prevent unauthorized devices from accessing.

3. The method of claim 1, wherein, The student terminal of the electronic paper book is based on electronic ink screen and has a system level application whitelist mechanism, which shields video playback, game download, social software and other non-learning functions, and only retains homework tutoring, reading and annotation functions authorized by the parent, thereby realizing physical and system double-layer control.

4. The method of claim 1, wherein, The parent terminal has a real-time mirror viewing function of the student terminal learning process, which can release the answer, generate an explanation substitute or stop operation at any time during the AI explanation process.

5. The method of claim 1, wherein, The abnormal bypass lock includes detecting whether the student terminal skips the strategy template to directly request the answer, uploads pictures irrelevant to learning, calls unlicensed model services or tries to access closed system resources.

6. The method of claim 1, wherein, The behavior chain record includes the question content of the student terminal, the model calling path, the parent approval record, the lock event, the system resource access attempt and the timestamp information, which are used for auditing and backtracking.

7. The method of claim 1, wherein, The generation of the parent teaching plan is initiated by the parent terminal, the system automatically labels the grade, subject, question type and label, and is only shared and referenced by the parent terminal, and the student terminal can only call in a controlled interface.

8. The method of claim 1, wherein, The auxiliary terminal includes but is not limited to a smart doll, a headset or a sound box, which can only play the explanation content that has obtained the "playable label" and refuses to play unauthorized or beyond the scope set by the parent.

9. The method of claim 1, wherein, The AI model calling does not limit the model type, brand or implementation, and the system only controls the permission path before and after calling, the compliance of output content and the closed running environment of the electronic paper book terminal.

10. The method of claim 1, wherein, The student end calls the whole behavior chain record of AI, the approval chain record and the teaching plan precipitation data are stored and audited with the parents as the responsibility attribution, and the illegal operation of bypassing the storage and auditing mechanism is prevented relying on the system closed characteristics of the electric paper book.

Images