A learning tutoring method and system based on the analysis of students' homework writing process
By collecting and analyzing multidimensional dynamic data during students' homework writing process, learning disabilities can be identified and personalized tutoring can be provided, solving the problem of traditional teaching methods being unable to identify learning disabilities and achieving highly efficient learning tutoring results.
Patent Information
- Application Number
- CN202611530720.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-30
- Publication Date
- 2026-11-17
AI Technical Summary
In traditional teaching and homework correction, teachers and intelligent correction systems fail to effectively perceive the students' thinking paths in the problem-solving process, resulting in the inability to identify learning obstacles and provide targeted tutoring.
By collecting multi-dimensional dynamic data during students' homework writing process, including the three-dimensional coordinates of the pen tip, pen pressure, pen speed, pause duration, and correction actions, learning features are extracted and diagnosed. Machine learning models are used to identify the types of learning disabilities and match personalized tutoring strategies.
It enables accurate identification and personalized tutoring of students' learning disabilities, forming a closed-loop mechanism from process data collection to root cause correction, thereby improving teaching efficiency and learning outcomes.
Smart Images

Figure CN123118322A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of intelligent teaching, specifically involving a learning tutoring method and system based on the analysis of students' homework writing process. Background Technology
[0002] In traditional teaching and homework correction scenarios, teachers and existing intelligent correction systems mainly focus on the correctness of students' final answers, while lacking effective perception of the students' thought processes during problem-solving. Students' writing behaviors such as corrections, pauses, and repeated erasures on their homework or test papers are often ignored as writing noise, or even simply attributed to poor writing habits.
[0003] However, the writing behavior of middle school students during problem-solving is not random noise, but rather an external representation of their internal cognitive state: frequent erasures may indicate a vague understanding of concepts, long pauses may reflect memory overload, and changes in the rewritten content after erasure directly expose points of mental stagnation. Traditional diagnostic methods that emphasize results over process prevent teachers from knowing where and why students are stuck, and can only provide generalized explanations based on incorrect answers or assign repetitive, rote practice. This is not only inefficient but also fails to fundamentally address students' cognitive obstacles.
[0004] In view of the aforementioned technical deficiencies, this application is hereby filed. Summary of the Invention
[0005] The purpose of this application is to provide a learning tutoring method and system based on the analysis of students' homework writing process. By collecting multi-dimensional dynamic data of students' handwriting during the homework writing process, the system can deeply mine data features to restore students' cognitive state and accurately attribute and tutor students' thinking block, decision hesitation and learning obstacles in the problem-solving process.
[0006] The technical solution of this application is:
[0007] A learning tutoring method based on the analysis of students' homework writing process includes:
[0008] Step 1: Collect multi-dimensional dynamic data of students' handwriting, including the three-dimensional coordinates of the pen tip, pen pressure, pen speed, pause duration, and correction actions;
[0009] Step 2: Extract learning features from the multidimensional dynamic data of students' handwriting in their assignments, including geometric features, temporal features, hesitation features, and semantic association features of corrections.
[0010] Step 3: Input the learning characteristics of students' homework into the learning disability diagnostic model to obtain the type of learning disability;
[0011] Step 4: Based on the type of learning disability in the student's assignment, match the corresponding tutoring strategy from the tutoring strategy library and provide tutoring to the student.
[0012] Optionally, in the above-mentioned learning tutoring method based on the analysis of students' homework writing process, in step one, the high-frequency sensor built into the smart pen or touch terminal is used to collect multi-dimensional dynamic data of students' homework handwriting at a millisecond-level sampling frequency.
[0013] Optionally, in the above-mentioned learning tutoring method based on the analysis of students' homework writing process, in step two, the geometric feature extraction specifically includes extracting stroke length, curvature change, correction coverage area, and geometric center offset of the correction area.
[0014] Temporal feature extraction specifically includes calculating the time interval distribution between adjacent strokes, the duration of a single correction action, and the number of times the same position is repeatedly written;
[0015] Hesitation feature extraction specifically involves calculating features based on pen tip pressure, pen stroke speed, pause duration, and correction actions.
[0016] ;
[0017] in,
[0018] This is a characteristic of hesitation.
[0019] Let V be the variance of the pen tip pressure. This serves as the reference value for pen tip pressure.
[0020] The duration of low pen stroke speed, This represents the total duration of normal writing.
[0021] The total duration of the pause. The total writing time;
[0022] The number of times the correction was made. The total number of strokes in the writing;
[0023] , , , Weighting factors are used to calculate pen pressure, pen speed, pause duration, and correction action.
[0024] The semantic association feature extraction of the alteration is specifically carried out by using OCR / handwriting recognition technology to perform regional content recognition of the altered area, identify the causal chain of the content before the alteration, the alteration action, and the content after the alteration, and extract the differences between the content before and after the alteration.
[0025] Optionally, in the above-mentioned learning tutoring method based on the analysis of students' homework writing process, in step three, the learning disability diagnosis model adopts a random forest or deep neural network model to calculate and output the type of learning disability, including conceptual comprehension bias, learning memory overload, computational execution disorder, and problem-solving anxiety.
[0026] Optionally, in the above-mentioned learning tutoring method based on the analysis of students' homework writing process, in step four, for conceptual comprehension deviations, a concept tracing and reconstruction strategy is matched, and micro-lessons explaining knowledge points, analogy examples, and variation exercises are pushed to students.
[0027] To address learning and memory overload, a step-by-step decomposition guidance strategy is employed, which breaks down questions into several sub-steps and provides step-by-step hints.
[0028] To address computational performance barriers, we provide targeted computational training strategies, offering students specific computational practice and verification strategy guidance.
[0029] To address problem-solving anxiety, we employ anxiety-relief psychological counseling strategies, providing students with relaxation guidance, adaptive difficulty adjustment, and positive feedback incentives.
[0030] To counter blind guessing, we offer strategies for developing good reading habits, providing students with a checklist for reading questions and tips for slow thinking.
[0031] A learning tutoring system based on the analysis of students' homework writing process, used to implement the aforementioned learning tutoring method based on the analysis of students' homework writing process, includes:
[0032] Dynamic data acquisition module: used to collect multi-dimensional dynamic data of students' handwriting in their homework, including the three-dimensional coordinates of the pen tip, pen pressure, pen speed, pause duration, and correction actions;
[0033] Learning Feature Extraction Module: Used to extract learning features from multidimensional dynamic data of students' handwriting in their assignments, including geometric features, temporal features, hesitation features, and semantic association features of corrections;
[0034] Learning Disability Type Diagnosis Module: This module is used to input the learning characteristics of students' assignments into the learning disability diagnosis model to obtain the type of learning disability.
[0035] Tutoring strategy matching module: Based on the learning disability type of student's assignments, it matches the corresponding tutoring strategy from the tutoring strategy library to provide tutoring to the student.
[0036] Optionally, in the learning tutoring system based on the analysis of students' homework writing process described above, the dynamic data acquisition module uses a high-frequency sensor built into a smart pen or touch terminal to collect multi-dimensional dynamic data of students' homework handwriting at a millisecond-level sampling frequency.
[0037] Optionally, in the learning tutoring system based on the analysis of students' homework writing process, the geometric feature extraction module specifically extracts stroke length, curvature change, correction coverage area, and geometric center offset of the correction area.
[0038] Temporal feature extraction specifically includes calculating the time interval distribution between adjacent strokes, the duration of a single correction action, and the number of times the same position is repeatedly written;
[0039] Hesitation feature extraction specifically involves calculating features based on pen tip pressure, pen stroke speed, pause duration, and correction actions.
[0040] ;
[0041] in,
[0042] This is a characteristic of hesitation.
[0043] Let V be the variance of the pen tip pressure. This serves as the reference value for pen tip pressure.
[0044] The duration of low pen stroke speed, This represents the total duration of normal writing.
[0045] The total duration of the pause. The total writing time;
[0046] The number of times the correction was made. The total number of strokes in the writing;
[0047] , , , Weighting factors are used to calculate pen pressure, pen speed, pause duration, and correction action.
[0048] The semantic association feature extraction of the alteration is specifically carried out by using OCR / handwriting recognition technology to perform regional content recognition of the altered area, identify the causal chain of the content before the alteration, the alteration action, and the content after the alteration, and extract the differences between the content before and after the alteration.
[0049] Optionally, in the learning tutoring system based on the analysis of students' homework writing process, the learning disability type diagnosis module adopts a random forest or deep neural network model to calculate and output the learning disability type, including conceptual comprehension bias, learning memory overload, computational execution disorder, and problem-solving anxiety.
[0050] Optionally, in the learning tutoring system based on the analysis of students' homework writing process, the tutoring strategy matching module matches the concept comprehension deviation, and pushes micro-lessons on knowledge points, analogy examples and variation exercises to students.
[0051] To address learning and memory overload, a step-by-step decomposition guidance strategy is employed, which breaks down questions into several sub-steps and provides step-by-step hints.
[0052] To address computational performance barriers, we provide targeted computational training strategies, offering students specific computational practice and verification strategy guidance.
[0053] To address problem-solving anxiety, we employ anxiety-relief psychological counseling strategies, providing students with relaxation guidance, adaptive difficulty adjustment, and positive feedback incentives.
[0054] To counter blind guessing, we offer strategies for developing good reading habits, providing students with a checklist for reading questions and tips for slow thinking.
[0055] This application has at least the following beneficial technical effects:
[0056] This paper presents a learning tutoring method and system based on the analysis of students' homework writing process. It transforms students' handwriting into high-value feature data for the diagnosis of learning disabilities, restores students' learning disabilities, establishes a precise mapping algorithm from dynamic handwriting features to learning disability types, realizes the identification and classification of students' learning disabilities, and then matches targeted tutoring strategies to form a closed-loop mechanism from process collection to root cause correction. Attached Figure Description
[0057] Figure 1 This is a schematic diagram of a learning tutoring method based on the analysis of students' homework writing process provided in an embodiment of this application;
[0058] Figure 2 This is a schematic diagram of a learning tutoring system based on the analysis of students' homework writing process provided in an embodiment of this application.
[0059] To better illustrate this embodiment, some content in the accompanying drawings may be omitted, enlarged, or reduced. They are for illustrative purposes only and should not be construed as limiting the scope of this application. Detailed Implementation
[0060] To make the technical solution and advantages of this application clearer, the technical solution of this application will be described in a clearer and more complete manner below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only some embodiments of this application, and are only used to explain this application, not to limit this application. It should be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, and other related parts can be referred to the general design.
[0061] Furthermore, unless otherwise defined, the technical or scientific terms used in this application description shall have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The word "comprising" as used in this application description indicates that the concept preceding the word encompasses the concepts listed following the word and their equivalents, without excluding other related concepts.
[0062] A learning tutoring method based on the analysis of students' homework writing process, such as Figure 1 As shown.
[0063] Step 1: Collect multi-dimensional dynamic data of students' handwriting in their assignments.
[0064] Using high-frequency sensors built into smart pens or touch terminals, multi-dimensional dynamic data of students' handwriting can be collected at a millisecond-level sampling frequency.
[0065] The smart pen has a built-in accelerometer, gyroscope, pressure sensor, and infrared positioning module with a sampling frequency of no less than 100Hz. It can collect the pen tip's three-dimensional coordinates, pressure value, and motion status data in real time, and is also equipped with an eraser end sensor to identify erasing actions.
[0066] Touch terminals include tablet computers or interactive whiteboards, which collect handwriting through a touch layer with a sampling frequency of no less than 120Hz.
[0067] Multidimensional dynamic data, including:
[0068] Three-dimensional coordinates of the pen tip;
[0069] Pen tip pressure: reflects changes in writing pressure;
[0070] Pen stroke speed: calculated from coordinate difference and time interval;
[0071] Pause duration: defined as the duration for which the pen tip leaves the paper and the duration exceeds the pause threshold;
[0072] Correction actions include overwriting (writing over the original stroke), partial erasing (using the eraser end of the smart pen or a touch terminal to recognize the erasing gesture), and complete erasure.
[0073] Step 2: Extract learning features from the multidimensional dynamic data of students' handwriting in their assignments, including geometric features, temporal features, hesitation features, and semantic association features of corrections.
[0074] Geometric feature extraction specifically involves extracting spatial features such as stroke length, curvature variation, area covered by corrections, and the offset of the geometric center of the correction area.
[0075] Temporal feature extraction specifically includes calculating temporal features such as the distribution of time intervals between adjacent strokes, the duration of a single correction action, and the number of times the same position is repeatedly written.
[0076] Hesitation feature extraction specifically involves calculating features based on pen tip pressure, pen stroke speed, pause duration, and correction actions.
[0077] ;
[0078] in,
[0079] This is a characteristic of hesitation.
[0080] Let V be the variance of the pen tip pressure. The reference value for pen tip pressure is the 95th percentile of the pen tip pressure value sequence during normal writing (excluding corrections and pauses) and the larger value in the range of the pen tip pressure sensor.
[0081] The duration of low pen stroke speed, This represents the total duration of normal writing.
[0082] The total duration of the pause. The total writing time;
[0083] The number of times the correction was made. The total number of strokes in the writing;
[0084] , , , The weighting factors calculated for pen tip pressure, pen stroke speed, pause duration, and correction action are summed to 1. This can be initially set. , , , .
[0085] The semantic association feature extraction of alteration specifically involves using OCR / handwriting recognition technology to perform regional content recognition on the altered area, identify the causal chain of the content before alteration, the alteration action, and the content after alteration, and extract the differences between the content before and after alteration as direct evidence of points of conceptual confusion.
[0086] Step 3: Input the learning characteristics of students' homework into the learning disability diagnostic model to obtain the type of learning disability.
[0087] Learning disability diagnostic models can use machine learning models such as random forests or deep neural networks to calculate and output the type of learning disability.
[0088] Types of learning disabilities include:
[0089] Conceptual comprehension bias: This manifests as different or incorrect variations of the same knowledge point before and after the correction, with moderate hesitation but high frequency of corrections;
[0090] Learning and memory overload: manifested as a generally slow writing speed, frequent and scattered pauses, large area of corrections, and a high score for hesitation characteristics;
[0091] Calculation execution disorder: manifested as dense localized erasures at the level of numbers and symbols, the rewritten content after erasures still contains calculation errors, and drastic fluctuations in pen tip pressure;
[0092] Problem-solving anxiety manifests as significantly high pen pressure, fluctuating writing speed, rapid and repetitive erasing actions, and unnecessary cross-out of entire paragraphs.
[0093] Blindly guessing: Characterized by unusually fast writing speed with almost no pauses and very few corrections, but the answers are wrong and the hesitation characteristic score is extremely low.
[0094] If the learning disability does not fall under any of the above categories, the learning disability diagnostic model will output "no learning disability." The learning disability diagnostic model can be trained using the following methods:
[0095] Collect multidimensional dynamic data on the handwriting of multiple students from different grade levels;
[0096] Learning features are extracted from the multidimensional dynamic data of students' handwriting in their assignments, and experts label the types of students' learning disabilities to construct a training dataset.
[0097] The learning disability diagnostic model is trained using the training dataset. Cross-validation is used to evaluate the model's accuracy, recall, and F1 score. Once the preset values required for use are met, the training is complete and the model is deployed.
[0098] During the subsequent use of the learning disability diagnostic model, new training data will be continuously collected, and incremental training of the model will be triggered periodically for continuous optimization.
[0099] During the training of the learning disability diagnostic model, weighting coefficients can be calculated for pen tip pressure, pen stroke speed, pause duration, and correction actions. , , , Dynamic optimization is performed to achieve the best performance of the learning disability diagnostic model.
[0100] Step 4: Based on the type of learning disability in the student's assignment, match the corresponding tutoring strategy from the tutoring strategy library and provide tutoring to the student.
[0101] For misunderstandings of concepts, a concept tracing and reconstruction strategy is used to provide students with micro-lessons explaining knowledge points, analogies, and variation exercises.
[0102] To address learning and memory overload, a step-by-step decomposition guidance strategy is employed, which breaks down questions into several sub-steps and provides step-by-step hints.
[0103] To address computational performance barriers, we provide targeted computational training strategies, offering students specific computational practice and verification strategy guidance.
[0104] To address problem-solving anxiety, we employ anxiety-relief psychological counseling strategies, providing students with relaxation guidance, adaptive difficulty adjustment, and positive feedback incentives.
[0105] To counter blind guessing, we offer strategies for developing good reading habits, providing students with a checklist for reading questions and tips for slow thinking.
[0106] After the tutoring strategy is implemented, multidimensional dynamic data of students' handwriting in subsequent assignments are continuously collected. The changes in characteristics before and after the implementation of the tutoring strategy are compared to evaluate the intervention effect and optimize the feature extraction of multidimensional dynamic data, learning disability diagnosis model, tutoring strategy, etc., to form a closed loop of continuous optimization.
[0107] A learning tutoring system based on the analysis of students' homework writing process is provided to implement the learning tutoring method based on the analysis of students' homework writing process disclosed in the above embodiments. The system includes a dynamic data acquisition module, a learning feature extraction module, a learning disability type diagnosis module, and a tutoring strategy matching module. Figure 2 As shown.
[0108] The dynamic data acquisition module is used to collect multi-dimensional dynamic data of students' handwriting, including the three-dimensional coordinates of the pen tip, pen pressure, pen speed, pause duration, and correction actions.
[0109] The learning feature extraction module is used to extract learning features from the multidimensional dynamic data of students' handwriting in their assignments, including geometric features, temporal features, hesitation features, and semantic association features of corrections.
[0110] The learning disability type diagnosis module is used to input the learning characteristics of students' assignments into the learning disability diagnosis model to obtain the learning disability type.
[0111] The tutoring strategy matching module is used to match corresponding tutoring strategies from the tutoring strategy library based on the type of learning disability in students' assignments, and to provide tutoring to students.
[0112] Those skilled in the art should also recognize that the various modules of the learning tutoring system based on the analysis of student homework writing process disclosed in the embodiments of this application can be implemented in electronic hardware, computer software, or a combination of both. In order to clearly illustrate the interchangeability of hardware and software, they are generally described in terms of function in this application. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can choose different methods to implement the described functions for each specific application and its actual constraints, but such implementation should not be considered to be beyond the scope of this application.
[0113] The technical solution of this application has been described in conjunction with the preferred embodiments shown in the accompanying drawings. Those skilled in the art should understand that the scope of protection of this application is obviously not limited to these specific embodiments. Without departing from the principles of this application, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of this application.
Claims
1. A learning tutoring method based on the analysis of students' homework writing process, characterized in that, include: Step 1: Collect multi-dimensional dynamic data of students' handwriting, including the three-dimensional coordinates of the pen tip, pen pressure, pen speed, pause duration, and correction actions; Step 2: Extract learning features from the multidimensional dynamic data of students' handwriting in their assignments, including geometric features, temporal features, hesitation features, and semantic association features of corrections. Step 3: Input the learning characteristics of students' homework into the learning disability diagnostic model to obtain the type of learning disability; Step 4: Based on the type of learning disability in the student's assignment, match the corresponding tutoring strategy from the tutoring strategy library and provide tutoring to the student.
2. The learning tutoring method based on the analysis of students' homework writing process according to claim 1, characterized in that, In step one, the high-frequency sensor built into the smart pen or touch terminal is used to collect multi-dimensional dynamic data of the student's handwriting at a millisecond sampling frequency.
3. The learning tutoring method based on the analysis of students' homework writing process according to claim 2, characterized in that, In step two, the geometric feature extraction specifically involves extracting stroke length, curvature change, area covered by the correction, and the offset of the geometric center of the correction area. Temporal feature extraction specifically includes calculating the time interval distribution between adjacent strokes, the duration of a single correction action, and the number of times the same position is repeatedly written; Hesitation feature extraction specifically involves calculating features based on pen tip pressure, pen stroke speed, pause duration, and correction actions. ; in, This is a characteristic of hesitation. Let V be the variance of the pen tip pressure. This is the reference value for pen tip pressure; The duration of low pen stroke speed, This represents the total duration of normal writing. The total duration of the pause. The total writing time; The number of times the correction was made. The total number of strokes in the writing; , , , Weighting factors are used to calculate pen pressure, pen speed, pause duration, and correction action. The semantic association feature extraction of the alteration is specifically carried out by using OCR / handwriting recognition technology to perform regional content recognition of the altered area, identify the causal chain of the content before the alteration, the alteration action, and the content after the alteration, and extract the differences between the content before and after the alteration.
4. The learning tutoring method based on the analysis of students' homework writing process according to claim 3, characterized in that, In step three, the learning disability diagnostic model uses a random forest or deep neural network model to calculate and output the type of learning disability, including conceptual comprehension bias, learning and memory overload, computational execution disorder, problem-solving anxiety, and blind guessing.
5. The learning tutoring method based on the analysis of students' homework writing process according to claim 4, characterized in that, In step four, for conceptual misunderstandings, a concept tracing and reconstruction strategy is used to provide students with micro-lessons explaining knowledge points, analogies, and variation exercises. To address learning and memory overload, a step-by-step decomposition guidance strategy is employed, which breaks down questions into several sub-steps and provides step-by-step hints. To address computational performance barriers, we provide targeted computational training strategies, offering students specific computational practice and verification strategy guidance. To address problem-solving anxiety, we employ anxiety-relief psychological counseling strategies, providing students with relaxation guidance, adaptive difficulty adjustment, and positive feedback incentives. To counter blind guessing, we offer strategies for developing good reading habits, providing students with a checklist for reading questions and tips for slow thinking.
6. A learning tutoring system based on the analysis of students' homework writing process, used to implement the learning tutoring method based on the analysis of students' homework writing process as described in claim 1, characterized in that, include: Dynamic data acquisition module: used to collect multi-dimensional dynamic data of students' handwriting in their homework, including the three-dimensional coordinates of the pen tip, pen pressure, pen speed, pause duration, and correction actions; Learning Feature Extraction Module: Used to extract learning features from multidimensional dynamic data of students' handwriting in their assignments, including geometric features, temporal features, hesitation features, and semantic association features of corrections; Learning Disability Type Diagnosis Module: This module is used to input the learning characteristics of students' assignments into the learning disability diagnosis model to obtain the type of learning disability. Tutoring strategy matching module: Based on the learning disability type of student's assignments, it matches the corresponding tutoring strategy from the tutoring strategy library to provide tutoring to the student.
7. The learning tutoring system based on student homework writing process analysis according to claim 6, characterized in that, In the dynamic data acquisition module, a high-frequency sensor built into a smart pen or touch terminal is used to collect multi-dimensional dynamic data of students' handwriting at a millisecond-level sampling frequency.
8. The learning tutoring system based on student homework writing process analysis according to claim 7, characterized in that, In the learning feature extraction module, geometric feature extraction specifically involves extracting stroke length, curvature change, tamper coverage area, and geometric center offset of the tamper region. Temporal feature extraction specifically includes calculating the time interval distribution between adjacent strokes, the duration of a single correction action, and the number of times the same position is repeatedly written; Hesitation feature extraction specifically involves calculating features based on pen tip pressure, pen stroke speed, pause duration, and correction actions. ; in, This is a characteristic of hesitation. Let V be the variance of the pen tip pressure. This is the reference value for pen tip pressure; The duration of low pen stroke speed, This represents the total duration of normal writing. The total duration of the pause. The total writing time; The number of times the correction was made. The total number of strokes in the writing; , , , Weighting factors are used to calculate pen pressure, pen speed, pause duration, and correction action. The semantic association feature extraction of the alteration is specifically carried out by using OCR / handwriting recognition technology to perform regional content recognition of the altered area, identify the causal chain of the content before the alteration, the alteration action, and the content after the alteration, and extract the differences between the content before and after the alteration.
9. The learning tutoring system based on student homework writing process analysis according to claim 8, characterized in that, In the learning disability type diagnosis module, the learning disability diagnosis model uses random forest or deep neural network model to calculate and output the type of learning disability, including conceptual comprehension bias, learning and memory overload, computational execution disorder, problem-solving anxiety, and blind guessing.
10. The learning tutoring system based on student homework writing process analysis according to claim 9, characterized in that, In the tutoring strategy matching module, for conceptual misunderstandings, a concept tracing and reconstruction strategy is matched, and micro-lessons explaining knowledge points, analogy examples, and variation exercises are pushed to students. To address learning and memory overload, a step-by-step decomposition guidance strategy is employed, which breaks down questions into several sub-steps and provides step-by-step hints. To address computational performance barriers, we provide targeted computational training strategies, offering students specific computational practice and verification strategy guidance. To address problem-solving anxiety, we employ anxiety-relief psychological counseling strategies, providing students with relaxation guidance, adaptive difficulty adjustment, and positive feedback incentives. To counter blind guessing, we offer strategies for developing good reading habits, providing students with a checklist for reading questions and tips for slow thinking.
