An interactive dynamic problem-solving visualization method and system based on an AI learning system

Abstract

The present application relates to the technical field of intelligent online education, in particular to an interactive dynamic problem solving visualization method and system based on an AI learning system, belonging to the technical field of intelligent education. The method comprises: obtaining multi-modal input data of a problem input by a user, performing semantic analysis and knowledge point matching on the problem by an AI learning system, generating a structured step-by-step problem solving path, converting the structured problem solving path into an interactive visual description script, generating a dynamic problem solving rendering sequence with interactive control points, loading the dynamic problem solving rendering sequence by a front-end rendering module, and visually displaying a step-by-step problem solving animation. The present application solves the problem that existing AI problem solving tools only output static text answers, the problem solving process is abstract and lacks process interaction, and students have difficulty understanding the deduction logic. The present application realizes dynamic visualization of the problem solving process, step-by-step interactive backtracking, and parameter changeable recalculation, improves the understanding depth of students on the problem solving logic and underlying knowledge points, and is suitable for online intelligent teaching aid scenarios of mathematics, physics and other subjects.

Description

Technical Field

[0001] This invention relates to the field of intelligent online education technology, specifically to an interactive dynamic problem-solving visualization method and system based on an AI learning system. Background Technology

[0002] Current AI-powered educational tools, after receiving a question, mostly output the complete solution steps and final answer in text form. The output content is static text, and students can only passively read the complete answer. The problem-solving derivation process is a black box, and students cannot see the dynamic changes of variables and graphical conditions as the problem-solving steps progress.

[0003] The existing technology has the following shortcomings: First, abstract formulas, geometric models, and physical processes rely solely on textual descriptions, which makes them difficult to imagine. For problems with long logical deduction chains, students cannot keep up with the pace of deduction. Second, it lacks interactive capabilities, as it cannot go back or jump to a specific step in the solution for careful study, nor can it modify the known conditions of the problem to observe the changes in the results. Third, the system can only output the standard answer and cannot capture the student's behavior during the problem-solving process, nor can it pinpoint where the student's understanding is hindered. The assessment of learning progress relies solely on the correctness of the answers, lacking process-dimensional data.

[0004] While some solutions can generate problem-solving animations, these are pre-rendered, fixed videos where users cannot intervene in the problem-solving process or modify parameters. This lack of interactivity fails to meet the needs of inquiry-based learning. Therefore, an AI-driven technical solution is needed that dynamically visualizes the problem-solving process and supports user interaction. We propose an interactive, dynamic problem-solving visualization method and system based on an AI learning system to address these issues. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the present invention aims to propose an interactive dynamic problem-solving visualization method and system based on an AI learning system. This method and system can transform AI-generated structured problem-solving paths into interactive dynamic visualizations, support step jumps and rollbacks, modify and recalculate problem parameters, collect interactive behaviors for learning analysis, and help students understand the derivation logic. This invention solves the problems mentioned in the background technology.

[0006] To achieve the above objectives, the present invention provides the following technical solution: An interactive, dynamic problem-solving visualization method based on an AI learning system includes the following steps: S1. Obtain multimodal input data of the question input by the user, wherein the multimodal input data includes at least one of text input, photographic image input, and formula input.

[0007] S2. Call the AI ​​learning system to parse and process the multimodal input data of the question, identify the question conditions, known parameters, and solution objectives, match and associate knowledge points in the knowledge base, and generate a structured step-by-step problem-solving path. The structured step-by-step problem-solving path includes multi-level problem-solving steps, logical constraints of each step, intermediate calculation variables, and knowledge point tags.

[0008] S3. Convert the structured step-by-step problem-solving path into an interactive and visual description script, and generate a dynamic problem-solving rendering sequence with interactive control points, including step jump controls, parameter modification controls, and replay / pause controls.

[0009] S4. The front-end rendering module loads the dynamic problem-solving rendering sequence, renders the problem-solving animation step by step according to the time sequence, and synchronously displays the text derivation, formula, and graphic model.

[0010] S5. Real-time capture of user-issued interactive commands through interactive control points, including step jump commands, rollback commands, and question parameter modification commands.

[0011] S6. Based on the interactive instructions, perform recalculation on the structured step-by-step problem-solving path, update intermediate calculation variables, and synchronously drive the front-end rendering module to update the dynamic problem-solving visualization screen.

[0012] S7. Collect all user interaction data, combine it with the completion status of problem-solving paths, and output the learning analysis feedback results.

[0013] As a preferred technical solution, the generation of structured step-by-step problem-solving paths in S2 further includes: identifying multiple problem-solving approaches, generating multiple parallel structured step-by-step problem-solving paths, and receiving user selection instructions to switch between different problem-solving paths and display the corresponding dynamic problem-solving rendering sequences.

[0014] As a preferred technical solution, the interactive visualization description script in S3 is an SVG / JS executable script, in which a status flag is bound to each problem-solving step, and different status flags correspond to different rendering states of the visualization screen.

[0015] As a preferred technical solution, the execution process of the question parameter modification instruction in S6 includes: receiving the question parameters modified by the user, the AI ​​learning system re-executes the complete problem-solving calculation based on the modified parameters, generating an updated structured step-by-step problem-solving path, and then driving the front end to re-render the entire set of dynamic problem-solving animations.

[0016] As a preferred technical solution, the interactive behavior data in step S7 includes step jump records, parameter modification records, dwell time, and number of step replays. Statistical analysis of the interactive behavior data is performed to locate the user's weak knowledge points and push corresponding practice question resources.

[0017] An interactive dynamic problem-solving visualization system based on an AI learning system, used to implement the method described in any one of claims 1-5, comprising: a multimodal input acquisition module, an AI learning and parsing engine, a visualization script generation module, a front-end interactive rendering module, an interactive instruction processing module, a recalculation module, and a learning analysis module.

[0018] The multimodal input acquisition module is used to collect multimodal input data of the questions input by the user and transmit it to the AI ​​learning and parsing engine. The AI ​​learning and parsing engine is connected to the multimodal input acquisition module, the visualization script generation module, and the recalculation module, respectively, and is used to parse the questions, match the knowledge base, and output a structured step-by-step problem-solving path.

[0019] The visualization script generation module, connected to the AI ​​learning and analysis engine, receives structured step-by-step problem-solving paths, converts them into dynamic problem-solving rendering sequences with interactive control points, and sends them to the front-end interactive rendering module. The front-end interactive rendering module loads the dynamic problem-solving rendering sequences, renders the dynamic problem-solving visualization sequentially, receives user operations to generate interactive commands, and sends them to the interactive command processing module. The interactive command processing module, connected to both the front-end interactive rendering module and the recalculation module, parses the interactive commands and sends calculation scheduling signals to the recalculation module. The recalculation module, connected to both the AI ​​learning and analysis engine and the front-end interactive rendering module, receives the scheduling signals, completes the recalculation of the problem-solving paths, and sends the updated problem-solving paths back to the front-end interactive rendering module. The learning analysis module, connected to the front-end interactive rendering module, collects user interaction behavior data and outputs learning analysis feedback results.

[0020] As a preferred technical solution, the AI ​​learning and parsing engine has a built-in subject knowledge base, a large model reasoning unit, and a problem-solving path splitting unit. The problem-solving path splitting unit is used to break down the complete problem-solving answer into multi-level sub-steps with dependencies and mark the variable dependencies between the steps.

[0021] As a preferred technical solution, the interactive control points are embedded in the visual interface output by the front-end interactive rendering module. The interactive control points include a step progress bar, a parameter editing pop-up window, an animation play / pause / rewind button, and a multi-solution switching button.

[0022] As a preferred technical solution, the recalculation module is equipped with a variable verification unit. The variable verification unit is used to verify whether the modified question parameters conform to the subject constraints. If the parameters are invalid, a parameter exception prompt is returned to the front-end interactive rendering module, and the recalculation is not performed.

[0023] As a preferred technical solution, an offline export module is also included. The offline export module is connected to the visualization script generation module and is used to package the dynamic problem-solving rendering sequence into an offline viewable file, supporting local playback of the dynamic problem-solving visualization process without the AI ​​server.

[0024] The AI-generated problem-solving results are broken down into structured, step-by-step solution paths and transformed into interactive, dynamically rendered sequences, moving beyond static text answers. Formulas, geometric figures, and physical processes are dynamically demonstrated along with the solution steps, lowering the understanding threshold for abstract problems. This addresses the technical issues of traditional AI problem-solving outputting static text and lacking intuitive derivation processes. Interactive control points are set up to support step jumps, rewinding, and recalculation after modifying problem parameters. Students can independently explore how changes in conditions affect the solution results, achieving inquiry-based learning. Simultaneously, user interaction data is collected to identify weak knowledge points from the perspective of the problem-solving process, rather than solely relying on correct or incorrect answers to assess learning progress, thus enhancing the personalization of intelligent teaching aids.

[0025] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention breaks down the problem-solving results output by AI into a structured step-by-step problem-solving path, transforming it into an interactive and dynamically rendered sequence. Instead of static text answers, it dynamically demonstrates formulas, geometric figures, and physical processes along with the problem-solving steps, lowering the understanding threshold of abstract problems and solving the technical problem that traditional AI problem-solving outputs static text and the derivation process is not intuitive.

[0026] 2. This invention sets up interactive control points to support step jumps, rewinding, and recalculation by modifying the problem parameters. Students can independently explore the changes in the solution results caused by changes in conditions, realizing inquiry-based learning. At the same time, it collects user interaction data to identify weak knowledge points from the perspective of the problem-solving process, rather than judging learning progress solely based on the correctness of answers, thereby improving the personalization of intelligent teaching aids. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 This is a block diagram of the overall system architecture of the present invention; Figure 2 This is a flowchart illustrating the overall process of the method of the present invention.

[0029] The diagram shows: 1. Multimodal input acquisition module; 2. AI learning and analysis engine; 3. Visualization script generation module; 4. Front-end interactive rendering module; 5. Interactive command processing module; 6. Recalculation module; 7. Learning analysis module; 8. Offline export module. Detailed Implementation

[0030] The embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. In the description of the present invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as limiting the present invention. In addition, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined with "first" and "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, and "several" means one or more, unless otherwise explicitly specified.

[0031] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection, an electrical connection, or a connection that allows for communication; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0032] In this invention, unless otherwise explicitly specified and limited, "above" or "below" the second feature can include direct contact between the first and second features, or contact between the first and second features through another feature between them. Furthermore, "above," "over," and "on top" of the second feature includes the first feature directly above or diagonally above the second feature, or simply indicates that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature includes the first feature directly below or diagonally below the second feature, or simply indicates that the first feature is at a lower horizontal level than the second feature.

[0033] The following disclosure provides many different embodiments or examples for implementing various structures of the invention. To simplify the disclosure, specific examples of components and arrangements are described below. These are merely examples and are not intended to limit the invention. Furthermore, reference numerals and / or letters may be repeated in different examples; such repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed. In addition, examples of various specific processes and materials are provided in this invention, but those skilled in the art will recognize the application of other processes and / or the use of other materials.

[0034] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0035] According to the appendix Figure 1-2 As shown, this embodiment of the invention provides an interactive dynamic problem-solving visualization method based on an AI learning system, including the following steps: S1. Obtain multimodal input data of the question input by the user. The multimodal input data includes at least one of text input, photo image input, and formula input.

[0036] In practice, the multimodal input data comes from client devices, which can be web pages, mobile apps, or tablet learning terminals. Text input involves users manually entering the question text, while image input involves the terminal's camera capturing images of paper exercises or test papers, including images of questions that have undergone image preprocessing, distortion correction, and noise reduction. Formula input supports LaTeX and visual formula editors, directly carrying standardized formula semantic information. The system performs preliminary cleaning on the collected raw input data, removing invalid and noisy data, before submitting it completely to the backend AI learning system for analysis.

[0037] S2. The AI ​​learning system is invoked to parse and process the multimodal input data of the problem, identify the problem conditions, known parameters, and solution objectives, match and associate knowledge points in the knowledge base, and generate a structured step-by-step problem-solving path. The structured step-by-step problem-solving path includes multi-level problem-solving steps, logical constraints of each step, intermediate calculation variables, and knowledge point tags. After receiving the multimodal problem data, the AI ​​learning system completes semantic understanding of the problem stem, distinguishes known conditions, limiting conditions, and the object to be solved, extracts all problem parameters, calls the pre-built subject knowledge base to complete knowledge point association matching, and marks the concepts, theorems, and formulas covered by the problem. It further decomposes the complete problem-solving logic to form a machine-recognizable structured step-by-step problem-solving path. This structured step-by-step problem-solving path is not an ordinary text paragraph, but a structured dataset. Each level of problem-solving step is bound to the logical constraints that hold true for this step, the set of intermediate calculation variables generated by the step operation, and the corresponding knowledge point tags. At the same time, it records the data reference relationships between different steps, clarifies which preceding intermediate calculation variables subsequent steps depend on, and provides a data foundation for subsequent animation rendering and interactive recalculation.

[0038] As a further preferred technical solution of this method, the generation of structured step-by-step problem-solving paths in S2 also includes: identifying multiple problem-solving approaches, generating multiple parallel structured step-by-step problem-solving paths, and receiving user selection commands to switch between different dynamic problem-solving rendering sequences for display. For math, physics, and chemistry problems with multiple solutions, the AI ​​learning system identifies all feasible problem-solving approaches through large-scale model reasoning. For each problem-solving approach, it independently generates a complete and independent set of parallel structured step-by-step problem-solving paths. Multiple parallel problem-solving paths store their own problem-solving steps, logical constraints, calculation variables, and knowledge point tags, without interfering with each other. The system waits to receive user selection commands from the client, retrieves the corresponding problem-solving path based on the selection command, calls subsequent processes to generate the corresponding dynamic problem-solving rendering sequence, and achieves seamless switching between different problem-solving methods, making it convenient for students to compare multiple problem-solving approaches.

[0039] S3. Convert the structured step-by-step problem-solving path into an interactive visual description script, generating a dynamic problem-solving rendering sequence with interactive control points. The interactive control points include step jump controls, parameter modification controls, and replay / pause controls. The structured step-by-step problem-solving path output by the AI ​​learning system is backend business data and cannot be directly displayed on the front end. Therefore, it is necessary to translate the structured data into an interactive visual description script that the front end can parse and execute, and assemble it to generate a complete dynamic problem-solving rendering sequence. The rendering sequence not only contains the text, formulas, and graphic drawing instructions corresponding to each step, but also embeds the definition information of interactive control points, defining the control type, control trigger events, and the callback logic corresponding to the events. The interactive control points include at least step jump controls, parameter modification controls, and replay / pause controls to respond to different types of user operations.

[0040] As a further preferred technical solution of this method, the interactive visualization description script in S3 is an SVG / JS executable script. The script binds a status flag to each problem-solving step. Different status flags correspond to different rendering states of the visualization screen. The SVG / JS executable script is used as the visualization carrier. The SVG is responsible for drawing geometric figures, models, and formula layout, while the JS script is responsible for controlling the animation timing and interactive event response. The script binds a unique status flag to each problem-solving step. The status flags include at least three types: inactive, calculation in progress, and rendering complete. Different status flags correspond to different rendering states of the visualization screen: when the step is inactive, the corresponding graphic and formula elements are hidden; when the calculation is in progress, the corresponding elements are highlighted and flashing; and when rendering is complete, all derivation content is fully displayed. The status flags drive the screen changes, realizing the step-by-step animation effect of the problem-solving process.

[0041] S4. The front-end rendering module loads the dynamic problem-solving rendering sequence and renders the problem-solving animation step by step according to the time sequence. It simultaneously displays text derivations, formulas, and graphical models. The front-end rendering module receives and parses the dynamic problem-solving rendering sequence, reads the step order, status flags, and graphic drawing instructions in the script, and executes the rendering step by step according to the time sequence of the problem-solving logic. It does not output all the problem-solving results at once, but rather outputs the content step by step, imitating the derivation process on the whiteboard. The rendering process outputs three types of content simultaneously: text derivation text, mathematical / physical formulas, and geometric or physical simulation graphical models. As the problem-solving steps progress, it gradually draws graphics, expands formulas, and displays derivation text, intuitively presenting the graphical and numerical changes brought about by the changes in conditions, and transforming the abstract calculation logic into a visual dynamic screen.

[0042] S5. Real-time capture of user interaction commands issued through interactive control points. Interaction commands include step jump commands, backtrack commands, and problem parameter modification commands. The front-end rendering module continuously listens for user operation events triggered by the interface interactive control points and encapsulates user operations into standardized interaction commands. The interaction commands are mainly divided into three categories: step jump commands, which allow the user to directly locate a specific problem-solving step; backtrack commands, which allow the user to go back to the previous level of problem-solving steps; and problem parameter modification commands, which allow the user to modify the original known parameters in the problem stem. The front-end transmits the encapsulated interaction commands outward for subsequent modules to process.

[0043] S6. Based on the interactive instructions, the system performs recalculation on the structured step-by-step problem-solving path, updates the intermediate calculation variables, and synchronously drives the front-end rendering module to update the dynamic problem-solving visualization screen. After receiving the interactive instructions, the system performs recalculation on the structured step-by-step problem-solving path according to the instruction type, updates the internal intermediate calculation variable set, and then sends the updated data back to the front-end rendering module to drive the front-end to modify the status flags of each step, update the visualization screen, and respond to user interaction operations.

[0044] As a further preferred technical solution of this method, the execution process of the question parameter modification instruction in S6 includes: receiving the question parameters modified by the user; the AI ​​learning system re-executes the complete problem-solving calculation based on the modified parameters to generate an updated structured step-by-step problem-solving path; and then driving the front end to re-render the entire set of dynamic problem-solving animations. When the interaction instruction is a question parameter modification instruction, the system obtains the new question parameters edited by the user and sends the parameters back to the AI ​​learning system. The AI ​​learning system uses the modified question conditions to re-execute the question analysis and logical deduction completely, generating a new set of structured step-by-step problem-solving paths. Then, based on the new problem-solving paths, a new dynamic problem-solving rendering sequence is generated and sent to the front-end rendering module. The front end clears the original screen and completely re-renders the entire set of dynamic problem-solving animations. All intermediate variables, graphic models, and calculation results are updated synchronously with the new parameters, supporting students to explore the impact of changes in question conditions on the problem-solving results.

[0045] S7. Collect all user interaction data, combine it with the completion status of the problem-solving path, and output the learning analysis feedback results. During the playback of the problem-solving animation and the entire process of user interaction, continuously collect the raw data of various user operation behaviors, combine it with the execution and browsing completion status of the problem-solving path, perform statistics and analysis, and generate learning analysis feedback results. The feedback results can be displayed on the front-end page or saved to the user's learning file.

[0046] As a further preferred technical solution of this method, the S7 interactive behavior data includes step jump records, parameter modification records, dwell time, and number of step replays. Statistical analysis of the interactive behavior data identifies the user's weak knowledge points and pushes corresponding practice question resources. The collected interactive behavior data specifically includes: the user's step jump records, parameter modification records, page dwell time for each solution step, and number of times a single step is repeatedly replayed. It can also be extended to include records of switching between multiple solutions and animation pause / play records. The system performs statistical analysis on the above multi-dimensional behavioral data. For example, if the dwell time of a certain knowledge point-related step is much higher than the average level and it is replayed multiple times, then it is determined that the knowledge point belongs to the user's weak knowledge point. Based on the weak knowledge point identification results, corresponding practice question resources are pushed to the user to achieve personalized learning recommendations.

[0047] An interactive dynamic problem-solving visualization system based on an AI learning system, used to implement the method of any one of claims 1-5, comprising: a multimodal input acquisition module 1, an AI learning and parsing engine 2, a visualization script generation module 3, a front-end interactive rendering module 4, an interactive instruction processing module 5, a recalculation module 6, and a learning analysis module 7.

[0048] The multimodal input acquisition module 1 is used to collect multimodal input data of the questions entered by the user and transmit it to the AI ​​learning and parsing engine 2. The multimodal input acquisition module connects to various input components of the client, completes image preprocessing and format unification, and encapsulates different forms of input such as text, pictures, and formulas into standardized data messages and delivers the data to the backend.

[0049] The AI ​​learning and parsing engine 2 is connected to the multimodal input acquisition module 1, the visualization script generation module 3, and the recalculation module 6, respectively. It is used to parse questions, match knowledge bases, and output structured step-by-step problem-solving paths. As the core backend reasoning component of the entire system, the AI ​​learning and parsing engine receives the original question input, completes semantic parsing, retrieves knowledge bases to match knowledge points, and outputs structured step-by-step problem-solving paths carrying steps, logical constraints, calculation variables, and knowledge point tags. At the same time, it receives the modified question parameters returned by the recalculation module, completes re-reasoning calculations, and outputs the updated problem-solving path.

[0050] As a further optimized technical solution for this system, the AI ​​Learning and Analysis Engine 2 incorporates a subject knowledge base, a large model reasoning unit, and a problem-solving path decomposition unit. The problem-solving path decomposition unit breaks down the complete solution into multi-level sub-steps with dependencies, marking the variable dependencies between steps. The subject knowledge base stores theorems, formulas, and knowledge point metadata for each subject. The large model reasoning unit is responsible for problem understanding, logical deduction, and multiple solution generation. The problem-solving path decomposition unit receives the complete solution output by the large model, breaks down the coherent problem-solving process into multi-level sub-steps with data dependencies, explicitly marks the variable dependencies between steps, and clarifies which intermediate variables are called by subsequent steps, providing a basis for dependency for animation rendering and recalculation.

[0051] The visualization script generation module 3 connects to the AI ​​learning and analysis engine 2. It receives structured step-by-step problem-solving paths, converts them into dynamic problem-solving rendering sequences with interactive control points, and sends them to the front-end interactive rendering module 4. The visualization script generation module is responsible for translating backend structured business data into front-end executable scripts. It generates graphic, text, and formula rendering instructions for each step of the problem-solving content, embeds interactive control point definitions and step status markers, and assembles a complete dynamic problem-solving rendering sequence. This sequence is sent to the front-end and simultaneously pushed to the offline export module.

[0052] The front-end interactive rendering module 4 is used to load the dynamic problem-solving rendering sequence, render the dynamic problem-solving visualization screen in sequence, receive user operations, generate interactive instructions and send them to the interactive instruction processing module 5. The front-end interactive rendering module is deployed on the user terminal, parses the rendering sequence, completes the sequential playback of the problem-solving animation, and also carries the interface display of interactive control points. It captures user click, drag and edit operations, encapsulates user operations into standardized interactive instructions and sends them to the interactive instruction processing module. At the same time, it continuously collects user interaction behavior and reports it to the learning analysis module.

[0053] As a further preferred technical solution for this system, the interactive control points are embedded in the visual interface output by the front-end interactive rendering module 4. The interactive control points include a step progress bar, a parameter editing pop-up, an animation play / pause / replay button, and a multi-solution switching button. All interactive control points are directly integrated into the visual interface without the need to jump to other pages. The step progress bar enables step jumping, the parameter editing pop-up allows users to directly modify the known conditions of the problem, the animation play / pause / replay button controls the playback process of the problem-solving animation, and the multi-solution switching button is used to trigger the switching between multiple parallel problem-solving paths.

[0054] The interactive instruction processing module 5 is connected to the front-end interactive rendering module 4 and the recalculation module 6 respectively. It is used to parse interactive instructions and send calculation scheduling signals to the recalculation module 6. The interactive instruction processing module receives interactive instructions from the front end, identifies the instruction type, performs parameter verification preprocessing, generates corresponding calculation scheduling signals according to different instruction types, forwards the scheduling signals to the recalculation module, and controls the recalculation module to execute the corresponding processing logic.

[0055] The recalculation module 6 is connected to the AI ​​learning and analysis engine 2 and the front-end interactive rendering module 4 respectively. It is used to receive scheduling signals, complete the recalculation of the problem-solving path, and send the updated problem-solving path back to the front-end interactive rendering module 4. The recalculation module is the core scheduling unit of the interactive logic. For jump and rollback instructions, it completes the backtracking of internal variables of the original problem-solving path. For parameter modification instructions, it completes parameter verification. When necessary, it calls the AI ​​learning and analysis engine to recalculate and sends the updated problem-solving path back to the front-end interactive rendering module to drive the screen refresh.

[0056] As a further preferred technical solution for this system, the recalculation module 6 is equipped with a variable verification unit. The variable verification unit is used to verify whether the modified question parameters conform to the subject constraints. If the parameters are invalid, a parameter error message is returned to the front-end interactive rendering module 4, and the recalculation is not performed. The variable verification unit has built-in parameter constraint rules for the corresponding subject, such as geometric side length greater than 0 and physical quantity value range constraints. When the user modifies the question parameters, they are first checked by the variable verification unit. If the parameter type is incorrect or the value violates the objective constraints of the subject, a parameter error pop-up message is directly returned to the front end, and a recalculation request is not initiated to the back end, thus saving computing power and avoiding invalid calculations.

[0057] The learning analysis module 7 is connected to the front-end interactive rendering module 4. It collects user interaction behavior data, outputs learning analysis feedback results, receives all the raw data of interaction behavior reported by the front end, completes statistics and analysis, identifies users' knowledge weaknesses, generates learning reports, and outputs personalized practice push instructions.

[0058] As a further preferred technical solution for this system, an offline export module 8 is also included. This module is connected to the visualization script generation module 3 and is used to package the dynamic problem-solving rendering sequence into an offline viewable file, supporting local playback of the dynamic problem-solving visualization process without the need for an AI server. The offline export module receives the dynamic problem-solving rendering sequence output by the visualization script generation module, packages the script and rendering resources into an offline HTML file. After downloading and saving this file to a local device, it can be opened directly through a browser without connecting to the backend AI server, allowing for complete playback of the entire interactive dynamic problem-solving visualization process. This supports offline learning and the saving and sharing of course materials.

[0059] The interactive dynamic problem-solving visualization method and system based on an AI learning system of the present invention, when used: First, we collect and analyze the questions using multimodal methods: Users input questions into the terminal, which can be in the form of text, photographed images, or formulas. The original question data is then collected by the multimodal input acquisition module 1. This module is responsible for noise reduction, cropping, and distortion correction of images, standardizing the format of formulas, and uniformly encapsulating them into multimodal input data packets for transmission to the AI ​​learning and parsing engine 2.

[0060] AI Learning and Analysis Engine 2 is the core reasoning unit in the backend of this system. It integrates three major units: subject knowledge base, large model reasoning unit, and problem-solving path decomposition unit.

[0061] The large-scale model reasoning unit receives multimodal question data, completes semantic parsing of the question stem, identifies the question stem conditions, known parameters, and solution objectives, and retrieves the subject knowledge base to match the knowledge point tags corresponding to the question.

[0062] The problem-solving path decomposition unit breaks down the complete problem-solving logic output by the large model into multi-level sub-steps with dependencies, marks the variable dependencies between steps, and generates a structured step-by-step problem-solving path. This path is machine-readable structured data, containing multi-level problem-solving steps, logical constraints for each step, intermediate calculation variables, and knowledge point tags.

[0063] If a problem has multiple solutions, the large model reasoning unit will generate multiple independent parallel structured step-by-step solution paths for subsequent switching and invocation.

[0064] The AI ​​learning and analysis engine 2 outputs the generated structured step-by-step problem-solving path to the visualization script generation module 3. At the same time, the engine is bidirectionally connected to the recalculation module 6, receives the modified problem parameters passed in by the recalculation module 6, and sends back the updated problem-solving path after completing the recalculation.

[0065] Then, the visualization script is converted, the rendering sequence is generated, and the output is performed: The visualization script generation module 3 receives the structured step-by-step problem-solving path from the AI ​​learning and parsing engine 2, translates the backend structured business data into an executable SVG / JS visualization description script for the frontend, and binds a unique status flag [inactive, calculating, rendering complete] to each problem-solving step. Different status flags correspond to different rendering states of the frontend screen. The script embeds the definition of interactive control points and assembles and generates a dynamic problem-solving rendering sequence carrying interactive control points [step jump control, parameter modification control, replay pause control].

[0066] The dynamic problem-solving rendering sequence is divided into two outputs: The first transmission route sends data to the front-end interactive rendering module 4, and the second transmission route sends data to the offline export module 8. The offline export module 8 packages the entire dynamic problem-solving rendering sequence into an offline browsable file, an HTML file, which can complete the playback of problem-solving animations in a local browser without connecting to the AI server, realizing local courseware saving and offline learning.

[0067] The front-end interactive rendering module 4 is deployed on the user terminal, loads and parses the dynamic problem-solving rendering sequence, renders problem-solving animations step by step according to the time sequence of problem-solving logic, synchronously displays text deductions, formulas and graphic models to simulate the blackboard-writing style step-by-step deduction effect. The module interface embeds all interaction control points: a step progress bar, a parameter editing pop-up window, animation play / pause / playback buttons, and a multi-solution switching button. It receives user mouse and touch screen operations, encapsulates user operations into standardized interaction instructions, and sends them to the interaction instruction processing module 5.

[0068] Then, interaction instruction processing and recalculation are performed to update the screen, that is, the core interaction process: The interaction instruction processing module 5 receives the interaction instructions transmitted by the front-end interactive rendering module 4. The instruction types include a step jump instruction, a rollback instruction, and a problem setting parameter modification instruction. This module completes instruction parsing and preprocessing, generates a calculation scheduling signal, and issues it to the recalculation module 6.

[0069] The recalculation module 6 is the core of interactive calculation, with a built-in variable verification unit, which is respectively connected to the AI learning parsing engine 2 and the front-end interactive rendering module 4, and executes different processing logics according to the scheduling signal: When receiving a step jump / rollback instruction: the recalculation module 6 directly reads the original structured step-by-step problem-solving path, traces back to the intermediate calculation variable corresponding to the target step, returns the traced problem-solving path to the front-end interactive rendering module 4, and the front-end switches the screen rendering state according to the internal step state mark of the script, and directly jumps or rolls back to the target problem-solving step.

[0070] When receiving a problem setting parameter modification instruction: the built-in variable verification unit is first called to verify whether the parameter modified by the user meets the discipline constraints, for example, the side length of a geometry cannot be a negative number, and the value range constraint of physical quantities, If the parameter is invalid: a parameter abnormality prompt is returned to the front-end interactive rendering module 4, and no recalculation is performed to save computing power.

[0071] If the parameter is valid: the modified problem setting parameter is returned to the AI learning parsing engine 2, the AI learning parsing engine 2 completely re-parses and deduces based on the new parameter, generates an updated structured step-by-step problem-solving path and returns it to the recalculation module 6, the recalculation module 6 forwards the updated problem-solving path to the front-end interactive rendering module 4, and the front-end clears the original screen, completely re-renders the entire set of dynamic problem-solving animations, and all graphics, formulas and intermediate calculation results are updated synchronously following the new parameter.

[0072] The multi-solution switching logic is as follows: When the user clicks the multi-solution switching button on the front-end interactive rendering module 4 interface, a switching instruction is generated. The front-end loads the dynamic problem-solving rendering sequence corresponding to another parallel structured step-by-step problem-solving path, and completes the switching from one set of problem-solving animations to another set of solution animations.

[0073] Data collection and analysis of interactive behaviors: Throughout the entire process of the problem-solving animation playback and all user interactions, the front-end interactive rendering module 4 continuously reports user interaction behavior data to the learning analysis module 7. The collected interaction behavior data includes: step jump records, parameter modification records, dwell time at each step, number of steps reviewed, and records of switching between multiple solutions. The learning analysis module 7 combines the completion status of the problem-solving path browsing and statistically analyzes the interaction data. If it detects that the user reviews the steps bound to a certain knowledge point tag multiple times or spends too much time in those steps, it determines that the knowledge point is the user's weak knowledge point and finally outputs the learning analysis feedback results. On the one hand, it displays the learning report on the front end, and on the other hand, it pushes practice question resources for the corresponding knowledge point in a targeted manner, realizing personalized learning assessment based on the problem-solving process, rather than just judging the learning effect based on the correctness of the answers.

[0074] Finally, a complete data flow loop summary is provided: User inputs the question → Multimodal input acquisition module 1 collects data → AI learning and parsing engine 2 parses and generates a structured step-by-step problem-solving path. The problem-solving path is sent to the visualization script generation module 3 to generate a dynamic problem-solving rendering sequence with interactive control points. One path goes to the front-end interactive rendering module 4 for animation display, and the other path goes to the offline export module 8 to generate offline files.

[0075] User operations generate interactive commands on the front-end interface → Interactive command processing module 5 parses and generates scheduling signals → Recalculation module 6 processes them, and if necessary, calls back the AI ​​learning and parsing engine 2 to recalculate. The updated solution path is then returned to the front-end interactive rendering module 4 to refresh the visualization.

[0076] The front end continuously reports interactive behaviors → Learning analysis module 7 analyzes and outputs learning feedback.

[0077] It continuously responds to multiple user interactions, completing the entire interactive dynamic problem-solving visualization loop.

[0078] In the description of this specification, the references to terms such as "embodiment," "one implementation," "some implementations," "illustrative implementation," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with an implementation or example is included in at least one implementation or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same implementation or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more implementations or examples.

[0079] The basic principles, main features, and advantages of the present invention have been described above with reference to specific embodiments. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples of the present invention and are not intended to limit the invention. Various changes and modifications can be made to the present invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. An interactive dynamic problem-solving visualization method based on an AI learning system, characterized in that, Includes the following steps: S1. Obtain the multimodal input data of the question input by the user; S2. Call the AI ​​learning system to parse and process the multimodal input data of the question, identify the question conditions, known parameters, and solution objectives, match and associate knowledge points in the knowledge base, and generate a structured step-by-step problem-solving path. The structured step-by-step problem-solving path includes multi-level problem-solving steps, logical constraints of each step, intermediate calculation variables, and knowledge point tags. S3. Convert the structured step-by-step problem-solving path into an interactive and visual description script, and generate a dynamic problem-solving rendering sequence with interactive control points, including step jump control, parameter modification control, and replay pause control. S4. The front-end rendering module loads the dynamic problem-solving rendering sequence and renders the problem-solving animation step by step according to the time sequence, while simultaneously displaying text derivations, formulas, and graphical models. S5. Real-time capture of user-issued interactive commands through interactive control points, including step jump commands, rollback commands, and question parameter modification commands; S6. Recalculate the structured step-by-step problem-solving path according to the interactive instructions, update the intermediate calculation variables, and synchronously drive the front-end rendering module to update the dynamic problem-solving visualization screen. S7. Collect all user interaction data, combine it with the completion status of problem-solving paths, and output the learning analysis feedback results.

2. The interactive dynamic problem-solving visualization method based on an AI learning system according to claim 1, characterized in that: The generation of structured step-by-step problem-solving paths in S2 also includes: identifying multiple problem-solving approaches, generating multiple parallel structured step-by-step problem-solving paths, and receiving user selection commands to switch between different problem-solving paths and display the corresponding dynamic problem-solving rendering sequences.

3. The interactive dynamic problem-solving visualization method based on an AI learning system according to claim 1, characterized in that: The interactive visualization description script in S3 is an SVG / JS executable script. The script binds a status flag to each problem-solving step, and different status flags correspond to different rendering states of the visualization screen.

4. The interactive dynamic problem-solving visualization method based on an AI learning system according to claim 1, characterized in that: The execution process of the question parameter modification instruction in S6 includes: receiving the question parameters modified by the user, the AI ​​learning system re-executes the complete problem-solving calculation based on the modified parameters, generating an updated structured step-by-step problem-solving path, and then driving the front end to re-render the entire set of dynamic problem-solving animations.

5. The interactive dynamic problem-solving visualization method based on an AI learning system according to claim 1, characterized in that: The interactive behavior data mentioned in S7 includes step jump records, parameter modification records, dwell time, and number of step replays. Statistical analysis of the interactive behavior data is used to locate the user's weak knowledge points and push corresponding practice question resources.

6. An interactive dynamic problem-solving visualization system based on an AI learning system, used to implement the method described in any one of claims 1-5, characterized in that, include: Multimodal input acquisition module (1), AI learning and analysis engine (2), visualization script generation module (3), front-end interactive rendering module (4), interactive instruction processing module (5), recalculation module (6), learning analysis module (7); The multimodal input acquisition module (1) is used to acquire multimodal input data of questions input by users and transmit it to the AI ​​learning and parsing engine (2). The AI ​​learning and parsing engine (2) is connected to the multimodal input acquisition module (1), the visualization script generation module (3), and the recalculation module (6) respectively. It is used to parse questions, match knowledge base, and output structured step-by-step problem-solving path. The visualization script generation module (3) is connected to the AI ​​learning and analysis engine (2) and is used to receive the structured step-by-step problem-solving path, convert it into a dynamic problem-solving rendering sequence with interactive control points, and send it to the front-end interactive rendering module (4). The front-end interactive rendering module (4) is used to load the dynamic problem-solving rendering sequence, render the dynamic problem-solving visualization screen in sequence, receive user operation to generate interactive instructions and send them to the interactive instruction processing module (5). The interactive instruction processing module (5) is connected to the front-end interactive rendering module (4) and the recalculation module (6) respectively and is used to parse the interactive instructions and send the calculation scheduling signal to the recalculation module (6). The recalculation module (6) is connected to the AI ​​learning and analysis engine (2) and the front-end interactive rendering module (4) respectively and is used to receive the scheduling signal, complete the problem-solving path recalculation, and send the updated problem-solving path back to the front-end interactive rendering module (4). The learning analysis module (7) is connected to the front-end interactive rendering module (4), collects user interaction behavior data, and outputs the learning analysis feedback results.

7. The interactive dynamic problem-solving visualization system based on an AI learning system according to claim 6, characterized in that: The AI ​​learning and analysis engine (2) has a built-in subject knowledge base, a large model reasoning unit, and a problem-solving path splitting unit. The problem-solving path splitting unit is used to break down the complete problem-solving answer into multi-level sub-steps with dependencies and mark the variable dependencies between the steps.

8. The interactive dynamic problem-solving visualization system based on an AI learning system according to claim 6, characterized in that: The interactive control points are embedded in the visualization interface output by the front-end interactive rendering module (4). The interactive control points include a step progress bar, a parameter editing pop-up, an animation play / pause / replay button, and a multi-solution switching button.

9. The interactive dynamic problem-solving visualization system based on an AI learning system according to claim 6, characterized in that: The recalculation module (6) is equipped with a variable verification unit. The variable verification unit is used to verify whether the modified question parameters conform to the subject constraints. If the parameters are illegal, a parameter error message is returned to the front-end interactive rendering module (4), and the recalculation is not performed.

10. The interactive dynamic problem-solving visualization system based on an AI learning system according to claim 1, characterized in that: It also includes an offline export module (8), which is connected to the visualization script generation module (3) and is used to package the dynamic problem-solving rendering sequence into an offline viewable file, supporting local playback of the dynamic problem-solving visualization process without the AI ​​server.

Images