Intelligent projection method of teaching explanation content based on user identity and intelligent terminal

Abstract

This invention provides a user-identity-based intelligent projection method and intelligent terminal for teaching content, relating to the field of intelligent teaching technology. The method includes the following steps: first, acquiring the current user's identity information and collecting their real-time interactive behavior data on the projection interface; determining the demand baseline based on the identity information according to preset logic, determining the real-time cognitive feedback state based on the interactive behavior, and then combining both to match target teaching content from a teaching content library, adjusting the emphasis according to different user roles, and finally presenting it on the projection interface. This invention aims to solve the technical problems of existing intelligent projection devices using a "one-size-fits-all" approach, resulting in a mismatch between the teaching content and user needs, and inaccurate perception of user interaction intentions leading to poor tutoring targeting. It enables dynamic content adaptation based on user identity and cognitive state, achieving personalized teaching and precise tutoring.

Description

Technical Field

[0001] This invention relates to the field of intelligent teaching technology, and more specifically, to an intelligent projection method and intelligent terminal for teaching content based on user identity. Background Technology

[0002] In modern education, smart projectors, with their eye-friendly diffuse reflection imaging, are gradually becoming an important tool for assisting teaching. However, existing smart projectors still have significant limitations in educational applications. In actual teaching, users with different roles (such as students, parents, and teachers) have fundamentally different needs for teaching content, and the feedback from the same user varies at different cognitive stages. Existing technologies often adopt a "one-size-fits-all" approach, failing to dynamically adjust the focus and presentation of the content based on the user's identity characteristics and real-time interactive behavior.

[0003] Furthermore, during projection interaction, the system struggles to effectively distinguish between valid and disruptive user interactions, leading to inaccurate assessments of the user's cognitive state. When the projection interface is information-dense, the system often fails to accurately interpret the user's true intentions, resulting in a disconnect between interactive feedback and actual needs. Simultaneously, in multi-user interaction scenarios, the lack of effective permission management and smooth content switching mechanisms can easily cause chaos in the teaching process and gaps in the connection of knowledge points. These issues make it difficult for smart projectors to meet the complex and ever-changing educational needs in terms of personalized teaching, precise tutoring, and interactive experience. There is an urgent need for a projection solution that can deeply perceive user identity, accurately identify interactive intentions, and intelligently adapt content.

[0004] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0005] The purpose of this invention is to provide a smart projection method and smart terminal for teaching content based on user identity. It aims to solve the technical problems of existing smart projection devices adopting a "one-size-fits-all" approach, which leads to a mismatch between the teaching content and user needs, and inaccurate perception of user interaction intentions, resulting in poor tutoring targeting. The invention can achieve dynamic content adaptation based on user identity and cognitive state, realizing personalized teaching and precise tutoring.

[0006] In a first aspect, the present invention provides a method for intelligent projection of teaching content based on user identity, applied to a smart projector, the method comprising the following steps: S1. Obtain the identity information of the current user of the smart projector; S2. Collect real-time interactive behavior data of the current user on the projection interface; S3. Based on the identity information, determine the current user's needs according to preset logic; S4. Based on the real-time interactive behavior data, determine the current user's real-time cognitive feedback state; S5. Based on the demand benchmark and the real-time cognitive feedback status, target teaching content is matched from the teaching content library and presented on the projection interface; wherein, the target teaching content matched for different demand benchmarks has different focuses to adapt to users with different roles; The specific steps in step S2 include: S21. Obtain the saliency distribution map of the currently presented content in the projection interface; the saliency distribution map is used to characterize the visual attractiveness weight of different local areas in the projection interface; S22. By spatially mapping real-time interactive behavior data with a saliency distribution map, the effective interactive behaviors and disruptive behaviors generated by the current user can be distinguished and identified.

[0007] The intelligent projection method for teaching content based on user identity provided by this invention has the advantages of dynamically adjusting the focus and presentation of the teaching content according to the user's identity characteristics and real-time interactive behavior, realizing personalized teaching and precise tutoring, and effectively improving the access control and content switching smoothness in multi-user interaction scenarios.

[0008] In a second aspect, the present invention provides a smart terminal, comprising: An identity acquisition unit is used to acquire the identity information of the current user of the smart projector; The data acquisition unit is used to collect real-time interactive behavior data of the current user on the projection interface; The benchmark determination unit is used to determine the current user's demand benchmark based on the identity information and according to preset logic; A cognitive determination unit is used to determine the current user's real-time cognitive feedback state based on the real-time interactive behavior data. The matching display unit is used to match target teaching content from the teaching content library according to the demand benchmark and the real-time cognitive feedback status, and to present the target teaching content on the projection interface; wherein, the target teaching content matched for different demand benchmarks has different focuses to adapt to users with different roles.

[0009] As can be seen from the above, the intelligent projection method for teaching content based on user identity provided by this invention constructs a dual matching mechanism based on identity needs and real-time cognitive feedback by acquiring the user's identity information and real-time interactive behavior, overcoming the shortcomings of the "one-size-fits-all" explanation in existing technologies. By introducing saliency distribution maps and spatial mapping technology, interference can be effectively eliminated, and by combining time decay and spatial aggregation factors, accurate quantification of user attention and interactive intent can be achieved, solving the difficulty of intent parsing caused by misoperation in information-dense areas. At the same time, the permission management of exclusive learning sessions and the smooth switching mechanism based on knowledge anchors ensure the standardization of teaching activities and the coherence of knowledge points in multi-user scenarios, significantly improving the interactive experience and application accuracy of intelligent projectors in personalized tutoring.

[0010] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings. Attached Figure Description

[0011] Figure 1 This is a flowchart of a user-identity-based intelligent projection method for teaching content provided in an embodiment of the present invention.

[0012] Figure 2 This is a schematic diagram of a smart terminal provided in an embodiment of the present invention.

[0013] Label Explanation: 100. Identity Acquisition Unit; 200. Data Acquisition Unit; 300. Benchmark Determination Unit; 400. Cognition Determination Unit; 500. Matching and Display Unit; 600. Session Control Unit. Detailed Implementation

[0014] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0015] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0016] In modern educational practice, smart projectors, as teaching aids with diffuse reflection imaging characteristics, are widely used in homework correction and knowledge explanation scenarios. However, in actual application, the system struggles to provide accurate, tailored, and targeted teaching content based on the individual needs and real-time cognitive feedback of users with different roles, such as students, parents, and teachers. This problem manifests as a lack of deep analysis of user identity attributes when handling multi-role interactions, leading to a discrepancy between the teaching content delivery logic and the user's actual cognitive state, thus affecting the accuracy of teaching interaction and the effectiveness of information transmission.

[0017] For example, in homework tutoring scenarios, when students encounter incorrect math problems on the projected interface, the system often only provides standardized solution steps. If a student shows difficulty understanding a specific concept, the system cannot recognize prolonged lingering or repetitive actions on that area, making it unable to adjust the depth of explanation in a timely manner. Similarly, if parents intervene, the system fails to recognize their identity and switch to a more instructional guidance mode, making it difficult for parents to obtain targeted tutoring advice. In teacher-use scenarios, the system cannot adjust the focus of the explanation in real-time based on the teacher's gestures or voice prompts regarding specific content, resulting in a lack of personalized adaptation to different student levels. This information asymmetry in the interaction process prevents the system from dynamically adjusting the presentation of teaching content based on the user's real-time cognitive feedback, limiting the application depth of smart projectors in educational settings.

[0018] If the above problems are not addressed, the system will remain in a passive display state, unable to achieve real-time matching of teaching content with users' cognitive needs. This will not only lead to redundancy or omissions in teaching content, reducing the efficiency of users acquiring effective information, but also cause cognitive load for users during the learning process due to the lack of targeted interactive feedback. This will weaken the core value of smart projectors in assisting teaching, making it difficult for the system to meet the personalized teaching needs in complex educational scenarios.

[0019] For reference, see the appendix. Figure 1 This invention provides a user-identified intelligent projection method for teaching content, applied to a smart projector. The method includes the following steps: S1. Obtain the identity information of the current user of the smart projector; the identity information includes student role, parent role, and teacher role; S2. Collect real-time interactive behavior data of the current user on the projection interface; real-time interactive behavior data includes the duration of the current user's stay on a local area of ​​the projection interface, the number of times the user repeatedly operates on local content of the projection interface, the gestures on local content, and the question commands issued by the current user through voice. S3. Based on identity information and according to preset logic, determine the current user's needs baseline; S4. Based on real-time interactive behavior data, determine the current user's real-time cognitive feedback status; S5. Based on the demand baseline and real-time cognitive feedback status, match the target teaching content from the teaching content library and present the target teaching content on the projection interface; among them, the target teaching content matched with different demand baselines has different focuses to adapt to users with different roles.

[0020] For ease of understanding, the following explains some key terms in this embodiment: Smart projector: A device that integrates projection, computing, and interactive functions, capable of projecting digital content onto a screen or wall and supporting user interaction through gestures, voice, or other means. In this application, smart projector specifically refers to a device with learning assistance functions, capable of providing personalized teaching explanations based on user behavior and identity.

[0021] Identity information: refers to data used to identify the role attributes of the current user of the smart projector. In this application, identity information includes student roles, parent roles, and teacher roles, which determine the user's focus and level of understanding regarding the teaching content.

[0022] Real-time interactive behavior data: This refers to various behavioral data collected by the system in real time when the user interacts with the projection interface. This data reflects the user's attention to, understanding of, and questions regarding the projected content, such as dwell time, number of repeated operations, gestures, and voice commands.

[0023] Demand baseline: This refers to the standard of demand for teaching and explanation content determined based on user identity information and according to preset logic. Different roles correspond to different demand baselines; for example, students may need explanations of basic concepts, parents may need tutoring strategies, and teachers may need instructional analysis.

[0024] Real-time cognitive feedback status: This refers to the user's current level of understanding and mastery of the projected content, determined based on real-time interactive behavior data. This status is dynamically changing and reflects the user's real-time cognitive state during the learning process, such as attention span and level of confusion.

[0025] Teaching Content Library: A database storing various teaching content organized according to different knowledge points, difficulty levels, and teaching focuses. The system will match the most suitable target teaching content from this library based on demand benchmarks and real-time cognitive feedback.

[0026] Projection interface: refers to the display area projected by the smart projector, where users can view teaching content and perform interactive operations.

[0027] This method aims to address the challenge of providing accurate, tailored, and targeted teaching content to users with different roles in smart projector teaching scenarios, based on their individual needs and real-time cognitive feedback.

[0028] First, the method includes step S1, obtaining the identity information of the current user of the smart projector. The identity information includes student, parent, and teacher roles. There are several ways to obtain user identity information. For example, it can be done through user login information identification. The user logs in to the smart projector using a preset account and password. The system's internal authentication module verifies the entered credentials and extracts the role information associated with the account from the user database. Another method is to preset user profile selection. On the smart projector's interactive interface, the system projects multiple user avatars or name options. The user can select the corresponding profile through gesture recognition or a remote control, and the system loads the corresponding role information based on the selected profile. Additionally, it can be done through voice command recognition. The smart projector's built-in microphone continuously listens for preset voice commands. When the user says keywords such as "I am a student," "parent mode," or "teacher mode," the projector's internal voice processing unit captures and parses these commands to determine the current operator's identity.

[0029] Secondly, the method includes step S2, which collects real-time interactive behavior data of the current user on the projection interface. This real-time interactive behavior data includes the duration of the user's focus on a specific area of ​​the projection interface, the number of times the user repeatedly interacts with specific content on the projection interface, gestures related to specific content, and voice commands expressing questions. The high-speed camera of the smart projector captures the user's hand movements, gaze locations, and other operations within the projection area in real time. These interactive cues are processed by the system's internal interactive behavior analysis module. For example, when a user's gestures or gaze linger on an incorrect question or knowledge point for an extended period, the system starts a timer. If the dwell time exceeds a preset threshold, it indicates that the user may have questions about that knowledge point. If the user repeatedly clicks or gestures towards a correction prompt area, the system records the number of times this action is repeated. When the user zooms in or clicks on the projected explanation content using gestures or voice commands, the system records these actions, indicating that the user needs a deeper understanding of that content. Users can also express questions to the system via voice or gestures, and the voice processing unit will recognize these question commands.

[0030] Next, the method includes step S3, which determines the current user's needs baseline based on identity information and according to preset logic. Once the user's identity is identified, the system immediately retrieves preset learning preferences and cognitive characteristics related to that identity. This information is typically stored in a structured dataset. For example, a student's profile might include the student's grade level, subjects studied, common error types in past assignments, and preferences for different teaching methods. A parent's profile might include the parents' educational background, their focus on their child's learning, and tutoring experience. A teacher's profile might include the grade level taught, subjects taught, teaching focus, and areas of concern for common problems among students. This profile information is gradually improved during the user's initial registration or subsequent use and serves as the basis for determining the needs baseline.

[0031] Then, the method includes step S4, determining the current user's real-time cognitive feedback state based on real-time interactive behavior data. The system continuously observes the user's real-time interactive behavior on the current assignment grading interface and judges their current level of understanding and focus based on this behavioral data. For example, by analyzing dwell time, number of repeated operations, gestures, and question commands, the system can assess the user's concentration on specific knowledge points, level of confusion, and acceptance of the explained content, thereby forming a real-time cognitive feedback state.

[0032] Finally, the method includes step S5, which involves matching target explanation content from the teaching content library based on demand benchmarks and real-time cognitive feedback, and presenting the target explanation content on the projection interface. The target explanation content matched with different demand benchmarks has different focuses to adapt to users with different roles. Based on the identified user identity and real-time interaction behavior, the system dynamically determines the most suitable granularity of the explanation content. This is achieved through an intelligent decision-making logic that comprehensively considers identity information and interaction cues, selecting the most suitable explanation version from the preset explanation content library. For example, for students, if the system observes that a student pauses for a long time on a basic concept question or repeatedly checks the correction prompts, the system will determine that the student may not have a deep understanding of that basic knowledge point. In this case, the system will immediately select an explanation from the explanation content library that focuses on basic concepts, uses easy-to-understand language, and is supplemented with intuitive animations or diagrams. For parents, after identifying them, the system uses their pre-set educational background information and their attention shown on a complex question on the grading interface to determine if they need more in-depth background knowledge and tutoring methods. The system then projects content including the location of that knowledge point in the syllabus, analysis of common error types, and suggestions on how to guide the child's thinking. For teachers, based on their professional background and their interaction behavior of quickly browsing multiple student assignments and showing interest in a common error on the grading interface, the system determines if they need a more macro-level, analytical explanation. The system immediately projects an analysis of the teaching difficulties of that knowledge point, statistics on the frequency of that error among students, and a comparison of different problem-solving approaches.

[0033] The following example will provide a more detailed explanation of the above technical solution: Assume user A is a student using a smart projector to correct and explain math homework. In step S1, user A logs into the smart projector using their student ID and password, and the system identifies them as a "student." In step S2, user A sees the corrected result for a geometry problem on the projected interface. Their gesture lingers on a certain auxiliary line drawing step for 10 seconds, and they repeatedly point to that area while simultaneously issuing a voice command saying "I don't understand this." This behavioral data is collected by the system in real time. In step S3, based on user A's "student" identity information and their historical learning record (e.g., the student has a weak foundation in geometry), the system determines their need baseline to be "a detailed and intuitive explanation of basic concepts and problem-solving steps." In step S4, based on user A's real-time interactive behavioral data such as dwell time, repeated operations, and question commands, the system judges that they have significant cognitive difficulties with the auxiliary line drawing steps of the geometry problem, and their real-time cognitive feedback state is "confused but focused." In step S5, based on the requirement of "detailed and intuitive explanation of basic concepts and problem-solving steps" and the real-time cognitive feedback state of "confusion yet focused attention," the system matches an animated video explaining the drawing of auxiliary lines for the geometry problem from the teaching content library. This video dynamically demonstrates the process of generating auxiliary lines and provides easy-to-understand audio explanations. This targeted explanation is then presented on the projection interface to help user A understand the geometric concept.

[0034] In this way, smart projectors can adjust the depth, breadth, use of technical terms, and examples of the content in real time, ensuring that the content accurately matches the cognitive level and actual needs of different users. This avoids information overload or information shortage, significantly improves the effectiveness of the explanation, and ultimately enhances learning outcomes and teaching efficiency.

[0035] The aforementioned technical solution, by constructing a closed-loop intelligent interaction and content push mechanism, achieves a deep match between teaching content and user identity and cognitive state, thus making a significant contribution to existing technologies. Traditional intelligent learning devices often employ a uniform explanation mode when providing teaching content, making it difficult to offer personalized explanations for users with different roles (such as students, parents, and teachers), and also unable to dynamically adjust based on real-time cognitive feedback during the learning process. This results in potentially unsuitable content and low efficiency.

[0036] For example, in the above scenario, if a traditional method is used, when user A (the student) encounters confusion with a geometry problem, the system might only provide a standardized text or image explanation, instead of proactively matching an animated explanation video based on user A's "student role" and their real-time cognitive feedback state of "confusion and focused attention." This standardized explanation may not effectively resolve user A's specific confusion, leading to low learning efficiency.

[0037] By acquiring identity information, this application enables the system to establish differentiated teaching service benchmarks, ensuring that subsequent content delivery aligns with the specific learning objectives of each user. Collecting real-time interactive behavior data captures the distribution of user attention and points of confusion during the learning process, dynamically assessing the user's cognitive feedback status. Combining identity-based needs benchmarks with real-time cognitive feedback status allows for the selection of content from the content library that best matches the user's current learning progress and comprehension ability, presented with differentiated emphasis, effectively improving the relevance and interactive efficiency of teaching. This shift from passive display to proactive adaptation ensures that projected content accurately responds to the user's personalized learning needs, significantly improving the effectiveness of teaching and ultimately enhancing learning outcomes and teaching efficiency.

[0038] In some embodiments, the specific steps in step S2 include: S21. Obtain the saliency distribution map of the currently presented content in the projection interface; the saliency distribution map is used to characterize the visual attractiveness weight of different local areas in the projection interface; S22. By spatially mapping real-time interactive behavior data with a saliency distribution map, the effective interactive behaviors and disruptive behaviors generated by the current user can be distinguished and identified.

[0039] The process involves obtaining a saliency distribution map of the currently presented content on the projection interface. A saliency distribution map is an image processing result used to quantify the attractiveness of different regions in an image to human visual attention. Its function is to characterize the visual attractiveness weight of different local areas on the projection interface, i.e., to identify which areas are visual focal points or information priorities. One implementation method can use deep learning models, such as U-Net or SalGAN, to analyze the currently presented content on the projection interface and generate a pixel-level saliency map. Another implementation method can use traditional image processing algorithms, such as the ITTI model or frequency domain analysis, to analyze the currently presented content on the projection interface, extract visual features such as color, brightness, and edges, and then calculate and generate a saliency distribution map. Furthermore, semantic analysis can be combined; for example, for text content, titles, bold text, and key concept areas are assigned higher saliency weights; for images, faces and the central areas of objects are assigned higher saliency weights.

[0040] By spatially mapping real-time interactive behavior data to a saliency distribution map, valid and disruptive interactive behaviors generated by the current user can be distinguished. Real-time interactive behavior data originates from the dwell time, number of repeated operations, gestures, and voice commands collected in step S2 above, typically existing in the form of screen coordinates, timestamps, and operation types. Spatial mapping refers to associating real-time interactive behavior data (usually screen coordinates or regions) with corresponding regions on the saliency distribution map. In one implementation, the center coordinates of the interactive behavior or its covered area can be compared with pixel values ​​on the saliency distribution map. If the average saliency value of the interactive area is higher than a preset threshold, it is determined to be a valid interaction; otherwise, if the saliency value is low or zero, it is identified as disruptive behavior. In another implementation, an influence region can be defined on the saliency distribution map based on the type of interactive behavior (e.g., gesture pointing, click area), and statistical characteristics of the saliency values ​​within this influence region (e.g., maximum value, average value, cumulative value) can be calculated. These statistical characteristics are then compared with preset discrimination rules to distinguish between valid and disruptive interactive behaviors.

[0041] This application's solution quantifies the visual appeal weights of different areas within the projection interface by acquiring a saliency distribution map of the currently presented content, thereby identifying which areas are the focus or visual focal point of the teaching content. Based on this, by spatially mapping real-time interactive behavior data to the saliency distribution map, the system can determine whether the user's interaction occurs in areas with high visual appeal. This mapping mechanism correlates physical interaction coordinates with content-level visual weights, effectively distinguishing between effective interactive behaviors related to the teaching content and those that deviate from the teaching focus or are meaningless distractions. In this way, the system eliminates invalid noise, ensuring the purity of data for subsequent cognitive feedback state analysis, laying the foundation for more accurate personalized teaching content matching. This solution intelligently filters and accurately identifies the real-time interactive behavior data collected in step S2, making the determination of the user's real-time cognitive feedback state in subsequent step S4 more accurate, thereby improving the accuracy and personalization of target explanation content matching in step S5. This allows the entire intelligent projection method for teaching content to more effectively respond to the user's actual learning needs.

[0042] The following is a concrete example to illustrate this. Suppose a smart projector is projecting the solution to a math problem, and a student is interacting with the projected interface through gestures. First, the smart projector's image processing unit (e.g., a built-in graphics processor working with a pre-trained convolutional neural network model) analyzes the projected solution image and generates a saliency distribution map. This map identifies areas such as mathematical formulas, key calculation steps, and the final answer as high-saliency areas, while blank backgrounds or decorative elements are considered low-saliency areas. When a student points to an area on the projected interface, the smart projector's camera captures the real-time coordinates of the student's gesture. The system then spatially maps these gesture coordinates to the pre-generated saliency distribution map. If the student's gesture continuously points to a high-saliency mathematical formula area and remains there for more than a preset threshold, the system recognizes it as a valid interaction related to that formula. Conversely, if the student's gesture moves randomly in a blank area of ​​the projected interface or points to peripheral areas unrelated to the teaching content, the system identifies these as distracting behaviors because these areas have lower saliency weights. In this way, the system can accurately identify the knowledge points that students are truly interested in and ignore meaningless actions, thus providing high-quality interactive data for subsequent cognitive feedback analysis.

[0043] In some embodiments, when real-time interactive behavior data is spatially mapped to identify valid interactive behaviors that are ambiguous or imprecise, and the spatially mapped region of the valid interactive behavior is located in a densely populated local area of ​​the projected interface, the method further includes the following steps: S23. By parsing information-dense local regions, identify multiple independent semantic units contained within those local regions; S24. Based on the spatial characteristics of effective interactive behaviors, combined with the current user's identity information and learning preferences, calculate the correlation between effective interactive behaviors and multiple independent semantic units; S25. Based on the degree of relevance, determine the independent semantic unit to which the true intent of the effective interactive behavior is directed, and identify the effective interactive behavior as the precise interactive target pointing to the corresponding independent semantic unit.

[0044] In this context, "information-dense local areas" refer to visual areas on the projection interface that carry a large amount of independent information or knowledge points, and these information points are closely arranged in space. For example, teaching content may contain complex charts, formula derivations, or multiple parallel text explanations, all of which can constitute information-dense local areas. Alternatively, on a homework grading interface, displaying multiple information blocks simultaneously for a single question, such as the original question, student answers, standard answers, error analysis, and knowledge point links, with these information blocks being spatially adjacent, also falls under the category of information-dense local areas.

[0045] "Analyzing Information-Dense Local Areas" aims to conduct in-depth analysis of the visual content within these areas to identify the independent semantic units they contain. Specifically, image processing and Optical Character Recognition (OCR) technologies can be used to extract text, symbols, formulas, and other content from the local areas. Combined with Natural Language Processing (NLP) technology, semantic analysis is performed on the extracted content to identify independent knowledge points, concepts, or logical units. Alternatively, by pre-annotating the teaching content with a structured format, information-dense local areas can be divided into multiple independent units with semantic labels during content generation. The system can then directly read this pre-defined structured information during runtime.

[0046] "Multiple independent semantic units" refer to the smallest units of knowledge or information that can be independently understood and processed within a densely populated local area. For example, in mathematical formula derivation, each independent formula step, each variable definition, and each theorem reference can be considered an independent semantic unit. In reading comprehension, the main idea sentence of each paragraph, the explanation of each keyword, and the analysis of each rhetorical device can also be considered independent semantic units.

[0047] "Spatial characteristics of effective interactive behavior" refer to the spatial attributes of the projected interface area pointed to or covered by a user's gestures, gaze, or voice commands when performing effective interactive behavior, such as the geometry, size, center point, and boundaries. For example, the area pointed to by a gesture may be elliptical or rectangular, and its center point and coverage area can be precisely calculated. For voice commands, the keywords mentioned can be matched with the text content on the projected interface to determine the corresponding text area, and the coordinates and dimensions of this text area constitute its spatial characteristics.

[0048] "Identity information," as described above, includes student, parent, and teacher roles. These roles determine the user's depth of understanding of the teaching content, their focus, and their specific needs. "Learning preferences" refer to the personalized tendencies exhibited by users during the learning process, such as their mastery of specific knowledge points, their preference for different teaching formats (e.g., text, diagrams, videos), and their learning habits. Learning preferences can be obtained through historical learning data analysis, such as students' error rates on certain knowledge points, learning time, and preferred types of learning resources. They can also be collected through user selection in system settings or through questionnaires.

[0049] The goal of "calculating the correlation between effective interactive behaviors and multiple independent semantic units" is to evaluate the relevance or matching degree between effective interactive behaviors and various independent semantic units within an information-dense region. This can be achieved by calculating the geometric relationships such as the overlap and distance between the spatial features of the effective interactive behavior and the spatial locations of the independent semantic units, combined with semantic matching algorithms (e.g., if the interactive behavior is a voice command, the similarity between the speech recognition result and the text content of the semantic unit is calculated), and considering the weighting influence of user identity information and learning preferences on different semantic units, to comprehensively derive the correlation degree. Alternatively, a multimodal fusion model can be constructed, taking the spatial features of the interactive behavior, semantic information (such as the content of the voice command), user identity, and learning preferences as input, and outputting a correlation score with each independent semantic unit.

[0050] "Determining the independent semantic unit that the true intent of a valid interaction targets" means selecting the independent semantic unit with the highest relevance to the valid interaction based on the calculated relevance, as the target of the user's true intent. This can be done simply by selecting the semantic unit with the highest relevance, or by setting a relevance threshold, determining the target only when the highest relevance exceeds the threshold; otherwise, further prompting to clarify for the user may be necessary.

[0051] "Identifying effective interactive behaviors as precise interactive targets pointing to corresponding independent semantic units" refers to transforming originally vague or imprecise interactive behaviors into clear instructions that point to specific independent semantic units and can be precisely understood and responded to by the system through the above analysis process. For example, a gesture pointing to an area containing multiple formulas is, after processing, identified as an interaction precisely pointing to a specific formula. A vague voice question, "What does this mean?", is precisely understood as "The user has a question about the meaning of this independent semantic unit" after identifying the independent semantic unit it points to.

[0052] This application's solution effectively addresses the problem of ambiguous interactive intent recognition in complex information interfaces by introducing semantic parsing and correlation calculation mechanisms. In some of the aforementioned implementations, the system can distinguish between valid and distracting interactive behaviors generated by the current user. However, when these valid interactive behaviors themselves are ambiguous or imprecise, especially when they occur in information-dense local areas of the projected interface, the system still struggles to accurately determine the user's intent. This solution addresses this by first parsing the information-dense local areas, refining the macroscopic interactive area into multiple independent knowledge points or logical units, laying the foundation for accurate recognition. Second, by comprehensively considering the spatial characteristics of valid interactive behaviors, the user's identity information, and personalized learning preferences, the correlation between interactive behaviors and each semantic unit is calculated. This multi-dimensional correlation analysis filters out irrelevant interference and accurately captures the user's true learning intent. Finally, based on the correlation, the independent semantic units are determined, transforming ambiguous interactive behaviors into clear interactive goals, ensuring the relevance and accuracy of subsequent teaching content matching. This solution significantly enhances the interactive perception capabilities of smart projectors in complex teaching scenarios by elevating interactive behavior from the spatial mapping level to the semantic understanding level, enabling the system to respond more accurately to the personalized learning needs of different users.

[0053] The following example illustrates this. Suppose a complex physics problem is displayed on the projection interface of a smart projector, containing multiple formula derivations, concept definitions, and diagrams. These elements are densely packed on the screen, forming an information-dense local area. A student points to this local area, but the gesture covers two adjacent formula steps and one concept definition. Due to the large range of the gesture, the system initially recognizes it as a valid interaction, but the target is ambiguous. To resolve this ambiguity, the system first analyzes this information-dense local area. Through image recognition and text analysis techniques, the system identifies three independent semantic units within this area: formula A (Newton's Second Law), formula B (kinetic energy theorem), and concept C (definition of work). Next, based on the student's identity information (e.g., the student is a middle school student, and historical learning data shows a good grasp of Newton's Second Law but a shallow understanding of the kinetic energy theorem) and the spatial characteristics of the valid interaction (e.g., the gesture's center point is closer to formula B, and the coverage area has the highest overlap with the boundary of formula B), the system calculates the correlation between the gesture and formulas A, B, and C. The calculation results may show that the gesture has the highest correlation with formula B, followed by concept C, and the lowest correlation with formula A. Based on these correlations, the system determines that the student's true intention in making the gesture is towards formula B. Therefore, the system identifies this ambiguous gesture as a precise interactive target pointing to "formula B".

[0054] In some embodiments, the specific steps in step S4 include: S41. Eliminate distracting behaviors and calculate the current user's attention concentration on the current content in the projected interface based on the coverage weight of effective interaction behaviors in the saliency distribution map. S42. Determine the cognitive feedback state based on the level of attention concentration and the operational feedback corresponding to effective interactive behaviors.

[0055] Eliminating distracting behaviors refers to identifying and excluding user interactions on the projection interface that are irrelevant to understanding the current teaching content or learning objectives. For example, unconscious hand movements, brief gaze drifts, or accidental touches on non-teaching areas can all be considered distracting behaviors. Elimination methods can include setting behavior thresholds; for example, if the dwell time in a certain area is less than a preset minimum effective duration (e.g., 0.5 seconds), it is judged as a distracting behavior and eliminated. Alternatively, a machine learning model can be trained to identify and classify patterns irrelevant to the learning objectives, thereby automatically filtering out distracting behaviors. The coverage weight of effective interactive behaviors in the saliency distribution map refers to the degree of overlap and weight between the area where the effective interactive behavior occurs and the visually attractive areas in the saliency distribution map. The saliency distribution map is used to characterize the visual attractiveness weight of different local areas in the projection interface. For example, if an effective interactive behavior occurs in an area with a high weight in the saliency distribution map (such as core knowledge points or key annotations), its coverage weight is high; conversely, if it occurs in an area with a low weight (such as the background or decorative elements), its coverage weight is low. The coverage weight can be obtained by calculating the pixel-level or region-level weighted average of the effective interactive behavior area and the saliency distribution map. Calculating the current user's attention concentration on the current content in the projected interface is an indicator of the user's focus on the projected content. This calculation can be based on the coverage weight of effective interactive behaviors in the saliency distribution map. For example, the coverage weights of all effective interactive behaviors can be summed or weighted averaged to obtain a comprehensive attention score. Furthermore, factors such as the duration and intensity of effective interactive behaviors (e.g., the strength of gestures or the clarity of voice commands) can be combined to calculate attention concentration using a pre-defined algorithm model (e.g., a linear weighted model or a non-linear mapping function). The operational feedback corresponding to effective interactive behaviors refers to the system's response to the user's effective interactive behaviors. For example, when a user clicks on a knowledge point, the system may display a detailed explanation; when a user issues a question, the system may provide a voice answer. These operational feedbacks themselves, as well as the user's further interactions with these feedbacks (e.g., whether to click again, whether to continue asking questions), constitute operational feedback data. Determining the cognitive feedback state is the user's current state of understanding, confusion, mastery, or need for further explanation of the teaching content. Combining attention span and user feedback allows for a more accurate assessment of this state. For example, if attention span is high and user feedback indicates the user is deeply exploring a particular concept (e.g., repeatedly reviewing explanations or performing related exercises), the user may be in a "understanding" or "mastering" state. If attention span is high, but user feedback shows the user repeatedly issuing questioning commands or pausing for extended periods in the same area without progress, the user may be in a "confused" state. If attention span is low and user feedback is sparse or meaningless, the user may be in a "distracted" state.This determination can be achieved through a pre-defined rules engine or a machine learning-based classifier.

[0056] This application's solution introduces a saliency distribution map as a filtering mechanism, achieving precise screening and quantitative evaluation of user interaction behavior, thereby improving the reliability of cognitive feedback state judgment. First, the system eliminates distracting behaviors generated by the user on the projection interface. This process, based on the analysis of real-time interaction behavior data and combined with preset rules or models, separates non-targeted and meaningless actions from effective interaction behavior. This screening ensures that the data relied upon for subsequent cognitive evaluation is pure and relevant to the learning objectives, thus solving the problem of non-targeted actions interfering with cognitive evaluation. Based on this, the system calculates the current user's attention concentration on the current content on the projection interface according to the coverage weight of effective interaction behavior on the saliency distribution map. The saliency distribution map represents the visual attractiveness weight of different local areas on the projection interface, i.e., which areas are visual focal points or information priorities. Spatially mapping effective interaction behavior to the saliency distribution map allows identification whether the user is focusing on core teaching content with high visual appeal or interacting with peripheral or secondary content. This correlation makes the calculation of attention concentration more objective and accurate, quantifying the user's focus on core content. The system then combines the calculated attention concentration with the operational feedback corresponding to effective interactive behaviors to determine the user's cognitive feedback state. Operational feedback refers to the system's response to effective interactive behaviors and the user's further interaction with these responses. For example, a user's repeated clicking or zooming in on a knowledge point, or issuing question commands, and the user's subsequent behavior after the system provides an explanation. By comprehensively considering static attention metrics (attention concentration) and dynamic operational feedback, the system can construct a multi-dimensional evaluation model. For example, high attention concentration accompanied by in-depth exploration of the explained content may indicate understanding; high attention concentration accompanied by repeated questioning may indicate confusion; low attention concentration accompanied by sparse interaction may indicate distraction. This comprehensive evaluation allows the system to more accurately determine whether the user is currently in a state of understanding, confusion, or distraction, providing solid data support for the subsequent accurate matching of target explanation content.

[0057] The following example illustrates this. Suppose a smart projector is projecting a math problem's grading and explanation interface, which includes the original problem text, grading marks, a clickable "Solution Approach" area, and a "Knowledge Point Review" area. The smart projector's visual analysis module continuously collects real-time interactive behavior data from the user on the projected interface, such as gestures and dwell time. Simultaneously, the system has pre-generated a salience distribution map of the current grading and explanation interface. This map shows that the "Solution Approach" and "Knowledge Point Review" areas have high visual appeal weights, while the original problem text and grading marks areas have lower weights. When the user interacts with the projected interface, the system first eliminates distracting behaviors. For example, if the user's hand briefly swipes across the edge of the screen (less than 0.2 seconds) without pointing to any valid information area, this behavior is identified as distracting and ignored. If the user's gesture lingers in the "Solution Approach" area for more than 1 second, accompanied by a click, this behavior is identified as valid interaction. Subsequently, the system calculates the coverage weight of these valid interactions in the salience distribution map. For example, if a user clicks multiple times and lingers for an extended period in the "Solution Approach" area, the system assigns a high coverage weight to these interactions because this area has a high weight in the saliency distribution map. Conversely, if a user only briefly lingers in a minor corner of the problem text, even if it's recognized as a valid interaction, its coverage weight will be relatively low. The system then calculates the user's attention level regarding the math problem's explanation by weighting these coverage weights and combining them with the frequency and duration of the interaction. For instance, if a user frequently and for extended periods interacts with the "Solution Approach" and "Knowledge Point Review" areas, their attention level score will be high. Next, the system combines the calculated attention level with the operational feedback corresponding to the valid interactions to determine the user's cognitive feedback state. For example, if the attention level is high, and the user clicks on a sub-step within the "Solution Approach" area and then issues a voice command like "Why is this step like this?", this indicates that the user may be in a state of "confusion" and needs a more detailed explanation. If a user's attention span is high, and they quickly click "Next Question" or display a confident expression (via facial recognition) after reviewing the "Knowledge Point Review," it may indicate that they have "mastered" the knowledge point. Conversely, if their attention span is low, and they swipe randomly on the interface without a clear interaction goal, it may indicate that they are "distracted." The system can pre-set a set of rules, such as: if attention span is above a certain threshold and there are questioning commands, it is considered confused; if attention span is above another threshold and there are many repetitive actions targeting a specific area, it is considered to be in-depth understanding; if attention span is below a certain threshold and effective interaction is sparse, it is considered distracted.In this way, the system can dynamically and accurately determine the user's cognitive feedback state, providing a basis for subsequent intelligent matching of explanation content.

[0058] Through the above technical solution, this application effectively solves the problem of misjudgment of cognitive feedback state caused by mixed interfering behaviors in traditional methods. By first eliminating interfering behaviors, it ensures that the interaction data used for subsequent evaluation is pure and meaningful, thereby significantly improving data quality. Furthermore, by combining the coverage weight of effective interaction behaviors in the saliency distribution map to calculate attention concentration, the assessment of user focus is made more objective and accurate, distinguishing whether the user is focusing on core or peripheral content. On this basis, combining attention concentration with the operational feedback corresponding to effective interaction behaviors achieves a multi-dimensional and refined judgment of the user's cognitive state. This comprehensive evaluation mechanism enables smart projectors to more accurately identify whether the user is in a state of understanding, confusion, mastery, or distraction, thus providing a more reliable and accurate basis for matching target teaching content from the teaching content library. This not only avoids providing inappropriate teaching content due to misjudgment, but also enables teaching content to be truly personalized and intelligent, significantly improving teaching effectiveness and learning efficiency.

[0059] In some embodiments, when the current user generates multiple consecutive or parallel valid interactive behaviors on the projection interface, step S41, which calculates the current user's attention concentration on the current content on the projection interface based on the coverage weight of the valid interactive behaviors in the salience distribution map, includes the following specific steps: S411. Perform time series ranking and spatial proximity analysis on multiple effective interaction behaviors to obtain the time series ranking results and spatial proximity analysis results; S412. Based on the time series sorting results, apply a time decay factor to the coverage weights of multiple effective interaction behaviors in the saliency distribution map to obtain the coverage weights with applied time decay factors; the time decay factor decreases over time. S413. Based on the spatial proximity analysis results, a spatial aggregation factor is applied to the coverage weights of multiple effective interaction behaviors in the saliency distribution map to obtain the coverage weights with applied spatial aggregation factors; the spatial aggregation factor increases with the degree of proximity. S414. The coverage weights with applied time decay factors and applied spatial aggregation factors are weighted and superimposed to obtain the weighted superimposed coverage weights; S415. Based on the weighted overlay coverage weight, calculate the current user's attention concentration on the current content in the projected interface.

[0060] Specifically, time-series ranking and spatial proximity analysis are performed on multiple effective interactive behaviors to structure the complex interaction patterns of users on the projected interface from both temporal and spatial dimensions. Time-series ranking can be based on the timestamps of the interactive behaviors, for example, recording the precise start and end times of each gesture, voice command, or pause event and arranging them in chronological order. Spatial proximity analysis can employ clustering algorithms, such as K-means clustering or DBSCAN, to group spatially close interactive behaviors together, or to assess their proximity by calculating the geometric center distance of the regions where the interactive behaviors occur. These analytical results provide the foundational data for subsequent weight adjustments.

[0061] Applying a time decay factor based on the time series ranking results is essentially about acknowledging the immediate value of interactions. More recent interactions are more representative of the current cognitive state, while earlier interactions should have a gradually diminishing impact on current attention. The time decay factor can be an exponential decay function, such as f(t) = e^(-λt), where t is the time difference between the interaction and the current moment, and λ is the decay coefficient; or a linear decay function, such as f(t) = max(0, 1-kt), where k is the decay rate. This effectively filters out interference from past interactions on current attention evaluation, allowing attention to focus more on the user's current cognitive focus.

[0062] Applying a spatial aggregation factor based on spatial proximity analysis aims to identify the clustering effect of interactive behaviors. When multiple interactive behaviors are highly proximate in space, it often indicates that the user is focusing on a specific knowledge point or information area. The spatial aggregation factor can be a Gaussian kernel function, assigning higher weights to interactive behaviors in neighboring areas; or it can be an inverse distance weighting function, where the closer the interactive behavior, the larger its aggregation factor. By increasing the aggregation factor with increasing proximity, the visual attractiveness weight of local areas can be strengthened, thereby more accurately capturing the user's cognitive focus and reflecting their deep attention to specific content.

[0063] The coverage weights, which are subject to both a time decay factor and a spatial aggregation factor, are weighted and superimposed to comprehensively consider the timeliness and spatial aggregation of interactive behaviors. This weighted superposition can be achieved through a linear combination, for example, W_final = α * W_time_decay + β * W_spatial_aggregation, where α and β are adjustable weighting coefficients used to balance the influence of time and space factors on attention concentration. This superposition method allows the final calculated coverage weight to dynamically reflect the user's cognitive feedback during continuous interaction, effectively solving the problem of evaluation bias when multiple behaviors are parallel or continuous.

[0064] Based on the weighted overlay coverage weights, the user's current attention level on the current content in the projected interface is calculated. This can be achieved by normalizing the final weighted overlay coverage weights and combining them with a preset attention threshold model. For example, the normalized weight values ​​can be mapped to an attention score from 0 to 100, or attention can be judged by comparing the weight values ​​with a preset threshold.

[0065] This application's solution achieves refined evaluation of multiple interactive behaviors by introducing dynamic adjustment mechanisms in both temporal and spatial dimensions, thereby more accurately quantifying the user's attention concentration. By performing time-series ranking and spatial proximity analysis on multiple effective interactive behaviors, the complex interaction process can be deconstructed from both temporal and spatial dimensions, providing basic data support for subsequent weight adjustments. Applying a time decay factor based on the time-series ranking results acknowledges the immediate value of interactive behaviors; that is, the more recent the interactive behavior, the more representative it is of the current true cognitive state. The decay factor, which decreases over time, effectively filters out the interference of past behaviors on the current attention evaluation. Applying a spatial aggregation factor based on the spatial proximity analysis results identifies the clustering effect of interactive behaviors. When multiple interactive behaviors are highly proximate in space, it often means that the user is focusing on a specific knowledge point. The aggregation factor, which increases with the degree of proximity, strengthens the visual attractiveness weight of local areas, thereby more accurately capturing the user's cognitive focus. By weighting and superimposing the weights after time decay and spatial aggregation, the timeliness and spatial clustering are combined, so that the final calculated attention concentration can dynamically reflect the user's cognitive feedback during continuous interaction. This effectively solves the problem of evaluation bias when multiple behaviors are parallel or continuous, and provides a more reliable basis for matching personalized teaching content.

[0066] The following example illustrates this. Suppose a student is viewing a graded explanation of a math word problem on a smart projector. Within a short timeframe, the student first taps the keyword "speed" in the problem, then quickly moves their gesture to a formula area in the solution steps and holds it for 2 seconds. Immediately afterwards, they circle a calculation result below the formula. The system records these three consecutive valid interactions, noting their timestamps and spatial coordinates on the projected screen. First, the system sorts these three actions by time series and, based on their time difference from the current moment, applies different time decay factors to the coverage weight of each action in the saliency distribution map. For example, the highest weight is applied to the "circling the calculation result" action, and the lowest weight to the "tapping the keyword" action. Simultaneously, the system analyzes the spatial locations of the "holding the formula area" and "circling the calculation result" actions, finding them to be highly proximate. Therefore, based on their proximity, a higher spatial aggregation factor is applied to the coverage weights of these two actions. Finally, the system weights, after time decay and spatial aggregation processing, are weighted and summed to obtain a comprehensive weighted coverage weight. Based on this weighted summation of coverage weights, the system can more accurately calculate the student's attention concentration on the "formula application and calculation" part of the math word problem, thereby judging the student's cognitive feedback status on this knowledge point.

[0067] In some embodiments, step S1 is followed by: A1. Create a dedicated learning session for the current user, lock the learning session, and bind the interactive operation permissions of the learning session to the current user; A2. When an interaction is triggered by a user other than the current user, the interaction is intercepted, and an authorization transfer request is sent to the current user. At the same time, the system waits for the current user to respond with an authorization instruction or a denial instruction in response to the authorization transfer request. A3. If no authorization instruction is received within the preset response time, the default operation will be to refuse the instruction and maintain the locked state of the learning session.

[0068] A4. If an authorization instruction is received within the preset response time, the learning session status is updated by executing steps S2-S5 according to the new user's identity information, and the newly matched target explanation content is obtained. At the same time, before the newly matched target explanation content is presented on the projection interface, pre-generated transition guidance information is presented to achieve a smooth content switch.

[0069] Specifically, establishing and locking a dedicated learning session refers to the continuous process by which a smart projector provides personalized teaching and explanation services to a specific user. The system assigns each user a unique session ID and state space, ensuring that their learning progress, preference settings, and other information are stored and managed independently. Locking a session means that during session activation, the system marks the session as exclusive, preventing unauthorized operations from modifying or interrupting it. This can be achieved by maintaining a session state table in the system background, recording the ID and locking status of currently active sessions. Alternatively, when a session starts, the system allocates dedicated resources for that session in memory and sets access control flags, allowing only processes with the corresponding permissions to access and modify these resources. Binding the interactive operation permissions of the learning session to the current user means uniquely associating the right to operate the projection interface (such as gesture recognition, voice commands, remote control input, etc.) with the user currently using the smart projector. This means that only interactive commands issued by the current user will be responded to and processed by the system; operations by other users will be ignored or blocked. Implementation methods may include: after a user logs in or is identified, the system associates the user's unique identifier with the current session's permission token, and all subsequent interaction commands must be verified by the permission token; or, the system may use biometric technology (such as facial recognition or voiceprint recognition) to continuously verify the operator's identity, ensuring that only the current user can perform effective interaction.

[0070] When the system detects an interaction triggered by a user other than the current user, it means the system can detect an unauthorized user attempting to operate the projection interface or the smart projector itself. For example, the smart projector's camera can continuously monitor gestures within the projection area and use image recognition algorithms to determine if these gestures originate from the current authorized user. Additionally, the microphone can recognize voice commands and, combined with voiceprint recognition technology, determine if the voice originates from the current user. When an unauthorized user's operation is detected, the system triggers the corresponding processing flow. Intercepting interaction operations means the system prevents interaction commands issued by users other than the current user from being executed, thereby avoiding interference with the current learning session. This can be achieved by setting a permission verification layer in the system's input processing module. Any received interaction command must pass through this verification layer to verify whether its source is the current authorized user before being further processed. If verification fails, the command is discarded or marked as invalid. Sending an authorization transfer request to the current user means that when an unauthorized operation is detected, the system will prompt the current user to transfer control to another user. This request can be presented in various forms, such as a pop-up dialog box on the projection interface or a voice prompt. Waiting for the current user's authorization or denial instruction in response to the authorization transfer request means that after the system sends a request, it enters a waiting state to listen for the current user's feedback. The current user can respond to the request through preset interaction methods (such as clicking a button on the projection interface or issuing a voice command).

[0071] The preset response time refers to the maximum time the system waits for the current user's response after issuing an authorization transfer request. For example, it can be set to 5 seconds, 10 seconds, or longer. This time can be adjusted through system configuration. The default action of rejecting the request and maintaining the locked learning session means that if no explicit authorization instruction is received from the current user within the preset response time, the system will automatically reject the transfer and continue to maintain the locked learning session. This means that actions by unauthorized users remain ineffective, and the current user's learning process will not be interrupted.

[0072] Updating the learning session status by executing steps S2-S5 based on the new user's identity information means that after the current user explicitly authorizes the system, it obtains the new user's identity information and then re-executes steps S2 to S5 of the above method based on the new user's identity information. This means that the system will reassess the new user's needs and real-time cognitive feedback status based on the new user's identity and real-time interactive behavior, and match the most suitable target teaching content from the teaching content library. At the same time, the learning session status will also switch from the original user's status to the new user's status, including learning progress, preference settings, etc. Obtaining the newly matched target teaching content means that the system intelligently selects the teaching materials that best match the new user's current learning needs and cognitive level from the teaching content library based on the new user's identity information and real-time interactive behavior. Before presenting the newly matched target teaching content on the projection interface, pre-generated transition guidance information is presented first to avoid poor user experience or interruption of learning logic due to sudden content switching. The system will display a short, guiding message before displaying new content. The pre-generated transition guidance information aims to connect two different learning contents or session states. It can be a brief summary of the previous session and a preview or introduction of upcoming new content. Achieving smooth content transitions is achieved by presenting transitional guidance information. The system can effectively mitigate the abruptness caused by user switching content, helping new users quickly enter a learning state and maintaining the continuity of the learning process.

[0073] After obtaining the current user's identity information (S1), this method introduces a session management and access control mechanism. First, the system establishes a dedicated learning session for the current user and locks it, strictly binding all interactive operation permissions to that user. This means that during the session, only commands issued by the current user will be responded to by the system, ensuring the focus and continuity of the learning process. When the system detects that a non-user is attempting to trigger an interactive operation, such as through gestures or voice, it will not immediately respond to these operations but will first intercept these potential distractions. Subsequently, the system will send an authorization transfer request to the current user, asking whether to allow another user to take over the session. During this period, the system will wait for the current user's explicit feedback, i.e., an authorization instruction or a denial instruction. If the current user fails to provide a clear authorization instruction within a preset response time, the system will default to a denial operation and continue to maintain the locked state of the current learning session. This mechanism effectively prevents the session from being unexpectedly interrupted or control from being transferred without authorization due to the user's failure to respond in a timely manner, further ensuring the stability of the learning process. Conversely, if the current user explicitly issues an authorization command within the preset response time, the system will re-execute the core process of matching teaching content (S2-S5) based on the new user's identity information. This involves collecting the new user's real-time interactive behavior data, determining their needs and real-time cognitive feedback status, and matching the most suitable target teaching content from the teaching content library. Before presenting this new content on the projection interface, the system intelligently inserts a pre-generated transition guide. This transition guide aims to summarize the knowledge points of the previous session and warm up for the upcoming new session, thus achieving a smooth switch of teaching content and avoiding sudden changes in content and interruptions in learning logic caused by user identity switching. This greatly improves the learning experience and teaching efficiency in multi-user scenarios. Through the above-mentioned session locking, permission binding, interaction interception, authorization request, and smooth transition mechanisms, this method further solves the problems of interaction interference and unsmooth switching in multi-user environments, building upon the aforementioned user-identity-based personalized teaching method. It enables smart projectors to provide more stable, efficient, and humanized personalized teaching services in various multi-user education scenarios such as homes, tutoring institutions, or schools, ensuring that each user can have a focused and coherent learning experience in their respective learning sessions.

[0074] As a specific implementation method, a family scenario can be envisioned. The smart projector first identifies the current user as "student Xiaoming" through facial recognition or user login, establishes a dedicated learning session for him, locks the session, and binds all gesture and voice interaction permissions to Xiaoming. Xiaoming is reviewing and explaining his math homework on the projected interface. At this moment, Xiaoming's younger sister, Xiaohong, walks over and curiously touches the projected interface, attempting to operate it. The smart projector's camera captures Xiaohong's gesture and determines through image recognition that it is not Xiaoming's gesture. The system immediately intercepts Xiaohong's touch and pops up a prompt box on the projected interface, displaying "Unauthorized operation detected. Do you want to transfer control to Xiaohong? [Yes / No]", while simultaneously providing a voice prompt to Xiaoming. If Xiaoming does not click "Yes" or issue the "Authorize" voice command within 5 seconds, the system will automatically close the prompt box and maintain Xiaoming's locked learning session, rendering any actions by Xiaohong invalid. Xiaoming can then continue to focus on his math study. If Xiaoming clicks "Yes" or issues the "Authorize" voice command, the system will immediately obtain Xiaohong's identity information and reassess her learning needs based on that identity. The system will collect Xiaohong's real-time interactive behavior on the projection interface; for example, she might show interest in a cartoon or ask, "What is this?" Based on this information, the system will match new target content for Xiaohong from its teaching content library, such as a literacy animation for lower-grade students. Before presenting this literacy animation, the system will display a transitional guide on the projection interface. For example, the screen will first display the text, "Xiaoming has completed this math lesson; Xiaohong will now continue learning Chinese characters!" accompanied by a short animation effect, before smoothly switching to Xiaohong's literacy animation interface. In this way, both Xiaoming and Xiaohong can clearly understand the transition of the conversation, avoiding abrupt interruptions and ensuring the continuity of the learning process.

[0075] In some embodiments, transition boot information is generated through the following steps: B1. Extract the first knowledge anchor point of the learning session before the state update, and the second knowledge anchor point of the learning session after the state update, and generate transition guidance information connecting the first knowledge anchor point and the second knowledge anchor point; the transition guidance information is used to summarize the first knowledge anchor point and warm up the second knowledge anchor point.

[0076] Transitional guidance information refers to the information carriers used to connect previous and subsequent knowledge points and help users smoothly transition when switching teaching content. This information can be presented as a short text prompt summarizing the core points of the previous learning stage and introducing the learning theme of the next stage; or as an animation or diagram visually demonstrating the connection between knowledge points; or as an audio prompt reviewing the previous knowledge point and previewing the next; or it can be an interactive Q&A, guiding users to review old knowledge and think about new knowledge through questions. The first knowledge anchor point of the learning session before the status update refers to the core knowledge points, key concepts, or learning objectives of the current teaching content before the subject of the learning session changes or the content switches. It is the logical endpoint of the previous stage of learning content. The first knowledge anchor can be determined using Natural Language Processing (NLP) technology, automatically extracting keywords, key sentences, or thematic concepts from the currently presented teaching content; or it can be a tag pre-marked during the teaching content design, representing a stage of learning outcomes or core knowledge points; or it can be inferred by analyzing the user's interaction behavior (such as dwell areas and repetitive operations) in the last period before the switch, revealing the key knowledge points the user is focusing on. The second knowledge anchor after the learning session's status update refers to the starting knowledge point, introductory concept, or learning objective of the teaching content to be presented after the change in the subject or content of the learning session. It is the logical starting point for the next stage of learning content. This second knowledge anchor can be determined by analyzing the structure of the newly matched target teaching content, automatically identifying the core concepts or preparatory knowledge introduced at the beginning; or it can be a tag pre-marked during the teaching content design, representing the entry point or prerequisite for the new stage of learning; or it can be inferred by analyzing the new user's identity information and their needs, combined with the newly matched target teaching content, revealing the starting knowledge points the user is most likely to need. The purpose of generating transitional guidance information connecting the first and second knowledge anchors is to logically link and connect these two knowledge anchors, forming a coherent transition process. This can be achieved through pre-set templates, where the extracted first and second knowledge anchors are filled into template statements to form transitional text; or through knowledge graphs or semantic networks, analyzing the connection path between the two knowledge anchors and generating explanatory text or diagrams based on this path; or through machine learning models, learning and generating transitional content that effectively connects different knowledge points based on a large number of teaching switching cases. The transitional guidance information is used to summarize the first knowledge anchor and prepare for the second knowledge anchor. Its function is to summarize the previous knowledge point to consolidate learning outcomes, and to prepare for the next knowledge point to lower the learning threshold of new content, ensuring the continuity of learning. The summary can be a brief review of the first knowledge anchor and the extraction of core concepts; the preparation can be a background introduction to the second knowledge anchor, posing relevant questions, or stimulating thought.

[0077] This application's solution addresses the logical jump issue when switching teaching content between different users by introducing a transition guidance mechanism, achieving a smooth transition in the teaching process. When a user changes and authorization transfer is completed in a learning session, the system does not immediately present the new content when updating the learning session state and matching new target content. Instead, it first extracts the "first knowledge anchor" involved before the state update, which represents the logical core of the preceding learning content; simultaneously, it extracts the "second knowledge anchor" to be presented after the state update, which represents the entry point for subsequent learning content. The system then generates "transition guidance information" based on these two knowledge anchors. This "transition guidance information" is designed to summarize the "first knowledge anchor," helping users consolidate existing knowledge and avoid forgetting due to the switch; at the same time, it warms up the "second knowledge anchor," providing users with necessary background information and psychological preparation, making the presentation of subsequent content more coherent. This knowledge-anchor-based guidance method, within the overall framework of intelligent projectors dynamically adjusting the content to adapt to different user identities and real-time cognitive feedback states, especially in scenarios involving multi-user collaboration or handover, effectively avoids gaps in teaching logic and a fragmented learning experience by proactively building cognitive bridges before content switching, thus ensuring the integrity and systematic nature of the teaching explanation.

[0078] The following example illustrates this. Suppose student Xiaoming is learning a word problem about fraction addition and subtraction on a smart projector. Based on Xiaoming's identity and real-time interaction, the system is projecting an explanation and animation demonstrating the steps of "finding a common denominator for fractions with unlike denominators." At this point, the current learning session status is "Student Xiaoming, explanation of finding a common denominator for fractions with unlike denominators." When Xiaoming's parent, Ms. Li, wants to check on Xiaoming's learning progress and provide tutoring, she successfully takes over the learning session through an authorization transfer request (e.g., clicking the "Parent Takeover" button on the projection interface and entering a password). The system recognizes the new user as "Parent Ms. Li" and, based on her needs (e.g., parents may be more concerned about teaching methods for knowledge points and common mistakes their children make) and real-time cognitive feedback, matches new target explanation content, such as "Analysis of common errors in fraction word problems and tutoring strategies." To achieve a smooth content transition, the system first extracts the content Xiaoming was learning before the transition, identifying "finding a common denominator for fractions with unlike denominators" as the first knowledge anchor. Simultaneously, the system analyzes the "Common Errors in Fraction Word Problems and Tutoring Strategies" content to be presented to Ms. Li, identifying "Problem-solving strategies for fraction word problems" or "Teaching difficulties in the concept of fractions" as the second knowledge anchor. Based on these two anchors, the system generates transitional guidance information, such as displaying the text: "Xiaoming has just learned the method of finding a common denominator for fractions with unlike denominators, which is a key step in solving fraction addition and subtraction. Next, we will show you common error types and tutoring strategies for fraction word problems to help you better guide your child to understand the concept of fractions." Here, "Xiaoming has just learned the method of finding a common denominator for fractions with unlike denominators, which is a key step in solving fraction addition and subtraction" summarizes the first knowledge anchor, "finding a common denominator for fractions with unlike denominators"; while "Next, we will show you common error types and tutoring strategies for fraction word problems to help you better guide your child to understand the concept of fractions" serves as a warm-up for the second knowledge anchor, "Problem-solving strategies for fraction word problems" or "Teaching difficulties in the concept of fractions." Before formally presenting the "Common Errors in Fraction Word Problems and Tutoring Strategies," the smart projector displays the aforementioned transitional guidance information on the projection interface. By reading or listening to this transitional guidance information, Ms. Li was able to quickly understand Xiaoming's previous learning progress and prepare for the upcoming tutoring content, thus achieving a smooth transition in teaching content and avoiding the cognitive gap that might result from a direct change in content.

[0079] Reference Appendix Figure 2 This invention provides a smart terminal (which employs the user-identity-based intelligent projection method for teaching content described in the above embodiments; for details, please refer to the corresponding steps above), comprising: The identity acquisition unit 100 is used to acquire the identity information of the current user of the smart projector. Data acquisition unit 200 is used to collect real-time interactive behavior data of the current user on the projection interface; The benchmark determination unit 300 is used to determine the current user's demand benchmark based on identity information and according to preset logic; The cognitive determination unit 400 is used to determine the current user's real-time cognitive feedback state based on real-time interactive behavior data. The matching display unit 500 is used to match target teaching content from the teaching content library based on demand benchmarks and real-time cognitive feedback status, and present the target teaching content on the projection interface; among them, the target teaching content matched for different demand benchmarks has different focuses to adapt to users with different roles.

[0080] In some embodiments, a session control unit 600 is also included, which performs the following steps after the identity acquisition unit 100 has been activated: A1. Create a dedicated learning session for the current user, lock the learning session, and bind the interactive operation permissions of the learning session to the current user; A2. When an interaction is triggered by a user other than the current user, the interaction is intercepted, and an authorization transfer request is sent to the current user. At the same time, the system waits for the current user to respond with an authorization instruction or a denial instruction in response to the authorization transfer request. A3. If no authorization instruction is received within the preset response time, the default operation will be to refuse the instruction and maintain the locked state of the learning session.

[0081] A4. If an authorization instruction is received within the preset response time, the learning session status is updated by executing steps S2-S5 according to the new user's identity information, and the newly matched target explanation content is obtained. At the same time, before the newly matched target explanation content is presented on the projection interface, pre-generated transition guidance information is presented to achieve a smooth content switch.

[0082] In this context, the units described as separate components may or may not be physically separate. Similarly, the components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0083] Furthermore, the functional modules in the various embodiments of the present invention can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0084] In this document, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, without necessarily requiring or implying any such actual relationship or order between these entities or operations.

[0085] The above description is merely an embodiment of the present invention and is not intended to limit the scope of protection of the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A user-identity-based intelligent projection method for teaching content, applied to an intelligent projector, characterized in that, The method includes the following steps: S1. Obtain the identity information of the current user of the smart projector; S2. Collect real-time interactive behavior data of the current user on the projection interface; S3. Based on the identity information, determine the current user's needs according to preset logic; S4. Based on the real-time interactive behavior data, determine the current user's real-time cognitive feedback state; S5. Based on the demand benchmark and the real-time cognitive feedback status, target teaching content is matched from the teaching content library and presented on the projection interface; wherein, the target teaching content matched for different demand benchmarks has different focuses to adapt to users with different roles; The specific steps in step S2 include: S21. Obtain the saliency distribution map of the currently presented content in the projection interface; the saliency distribution map is used to characterize the visual attractiveness weight of different local areas in the projection interface; S22. By spatially mapping the real-time interactive behavior data with the saliency distribution map, the effective interactive behavior and interference behavior generated by the current user can be distinguished and identified.

2. The intelligent projection method for teaching content based on user identity as described in claim 1, characterized in that, The identity information includes student roles, parent roles, and teacher roles.

3. The intelligent projection method for teaching content based on user identity as described in claim 1, characterized in that, The real-time interactive behavior data includes the duration of the current user's stay on a local area of ​​the projection interface, the number of times the user repeatedly operates on local content in the projection interface, the gestures performed on the local content, and the question commands issued by the current user via voice.

4. The intelligent projection method for teaching content based on user identity as described in claim 1, characterized in that, When the real-time interactive behavior data, after spatial mapping, identifies valid interactive behaviors that are ambiguous or imprecise, and the spatially mapped region of the valid interactive behavior is located in a densely populated local area of ​​the projected interface, the method further includes the following steps: S23. By parsing the information-dense local region, identify multiple independent semantic units contained within the local region; S24. Based on the spatial characteristics of the effective interactive behavior, combined with the current user's identity information and learning preferences, calculate the correlation degree between the effective interactive behavior and the multiple independent semantic units; S25. Based on the correlation degree, determine the independent semantic unit to which the true intent of the effective interactive behavior points, and identify the effective interactive behavior as a precise interactive target pointing to the corresponding independent semantic unit.

5. The intelligent projection method for teaching content based on user identity as described in claim 1, characterized in that, The specific steps in step S4 include: S41. Eliminate the interfering behaviors and calculate the current user's attention concentration on the current content in the projection interface based on the coverage weight of the effective interaction behaviors in the saliency distribution map. S42. Determine the cognitive feedback state based on the attention concentration and the operation feedback corresponding to the effective interactive behavior.

6. The intelligent projection method for teaching content based on user identity as described in claim 5, characterized in that, When the current user generates multiple consecutive or parallel effective interactive behaviors on the projection interface, step S41, which calculates the current user's attention concentration on the current content in the projection interface based on the coverage weight of the effective interactive behaviors in the salience distribution map, includes the following specific steps: S411. Perform time series sorting and spatial proximity analysis on multiple effective interaction behaviors to obtain the time series sorting results and the spatial proximity analysis results; S412. Based on the time series sorting results, apply a time decay factor to the coverage weights of the multiple effective interaction behaviors in the saliency distribution map to obtain the coverage weights with applied time decay factors; the time decay factor decreases over time. S413. Based on the spatial proximity analysis results, a spatial aggregation factor is applied to the coverage weights of the multiple effective interaction behaviors in the saliency distribution map to obtain the coverage weights with applied spatial aggregation factors; the spatial aggregation factor increases with the degree of proximity. S414. The coverage weights of the applied time decay factor and the applied spatial aggregation factor are weighted and superimposed to obtain the weighted and superimposed coverage weights; S415. Based on the weighted overlay coverage weight, calculate the current user's attention concentration on the current content in the projected interface.

7. The intelligent projection method for teaching content based on user identity as described in claim 1, characterized in that, Step S1 is followed by: A1. Establish a dedicated learning session for the current user, lock the learning session, and bind the interactive operation permissions of the learning session to the current user; A2. When an interaction is triggered by a user other than the current user, the interaction is intercepted, and an authorization transfer request is sent to the current user. At the same time, the system waits for the current user to respond with an authorization instruction or a rejection instruction in response to the authorization transfer request. A3. If the authorization instruction is not received within the preset response time, a rejection operation will be performed by default and the learning session will remain locked. A4. If the authorization instruction is received within the preset response time, the state of the learning session is updated by executing steps S2-S5 according to the new user's identity information, and a new matching target explanation content is obtained. At the same time, before the new matching target explanation content is presented on the projection interface, pre-generated transition guidance information is presented to achieve a smooth content switching.

8. The intelligent projection method for teaching content based on user identity as described in claim 7, characterized in that, The transition guidance information is generated through the following steps: B1. Extract the first knowledge anchor point of the learning session before the state update and the second knowledge anchor point of the learning session after the state update, and generate transition guidance information connecting the first knowledge anchor point and the second knowledge anchor point; The transition guidance information is used to summarize the first knowledge anchor and warm up the second knowledge anchor.

9. A smart terminal, used in a smart projector, characterized in that, The method for performing the steps of the user-identity-based intelligent projection method for teaching content as described in claim 1 includes: An identity acquisition unit is used to acquire the identity information of the current user of the smart projector; The data acquisition unit is used to collect real-time interactive behavior data of the current user on the projection interface; The benchmark determination unit is used to determine the current user's demand benchmark based on the identity information and according to preset logic; A cognitive determination unit is used to determine the current user's real-time cognitive feedback state based on the real-time interactive behavior data. The matching display unit is used to match target teaching content from the teaching content library according to the demand benchmark and the real-time cognitive feedback status, and to present the target teaching content on the projection interface; wherein, the target teaching content matched for different demand benchmarks has different focuses to adapt to users with different roles.

Images