Intelligent question-driven personalized thinking training system

Abstract

The invention relates to an intelligent question-driven personalized thinking training system. The system comprises a data acquisition module, a guide strategy modeling module, a dynamic progressive question interaction module and a learning condition analysis and optimization suggestion module. The data acquisition module is used for acquiring guide data; a guidance strategy modeling module performs adaptability verification on the guidance data, and obtains a question style based on a verification result and a student background; vocabulary disassembling is carried out based on the teaching target, a mapping relation is established, and guide configuration data is acquired; the dynamic progressive question interaction module is combined with student input viewpoints, a question sub-module is used for generating a migration question of the teaching target, progressive question data and interaction state data are obtained, and the question data and the interaction state data are stored; the learning condition analysis and optimization suggestion module is used for analyzing the stored data and generating a structured evaluation report for guiding optimization suggestion; and personalized thinking training of students is realized.

Description

Technical Field

[0001] This invention belongs to the field of educational technology, specifically relating to a personalized thinking training system driven by intelligent questioning. Background Technology

[0002] Currently, intelligent thinking training still has the following areas that need improvement: In the current process of digital transformation in education, the cultivation of thinking skills has become one of the core teaching objectives, but existing thinking guidance technologies still have many problems that urgently need to be solved: Most teaching guidance relies on a didactic questioning model, where teachers dictate all questioning directions and students can only passively follow and respond. This lack of space for students to independently organize their thinking logic makes it difficult to stimulate their initiative in thinking, resulting in superficial thinking training that fails to truly improve students' logical reasoning and problem-solving abilities.

[0003] Current technologies lack a dynamic adjustment mechanism to match individual student differences. They fail to adjust the depth of guidance based on students' knowledge base and cognitive levels, nor do they consider the impact of interests and preferences on the effectiveness of guidance. This makes it easy for students with weak foundations to get stuck when faced with guidance beyond their capabilities, while advanced students cannot effectively improve due to insufficient guidance depth, creating a one-size-fits-all guidance dilemma. Question design lacks structured support, often consisting of scattered and random piles of questions, making it difficult to form a coherent thought-guiding path. In actual teaching, follow-up questions easily deviate from the preset teaching objectives, failing to effectively help students focus on core issues and engage in in-depth thinking, resulting in a low achievement rate of teaching objectives. The lack of real-time monitoring and targeted analysis of students' thinking performance during guidance makes it impossible to promptly identify and correct logical flaws and thinking deviations, nor can it provide teachers with comprehensive data on students' thinking development. Subsequent guidance optimization relies heavily on teachers' subjective experience and judgment, lacking scientific and quantitative basis, making it difficult to achieve continuous improvement in guidance quality.

[0004] With the increasing demand for personalized and precise education, traditional thinking guidance techniques can no longer meet the requirements of teaching in the new era. There is an urgent need for an intelligent system that can achieve proactive interactive guidance, dynamically adapt to individual differences, focus on teaching objectives, and provide quantitative feedback throughout the entire process. This system would overcome existing technological bottlenecks, promote the transformation of thinking training from passive reception to active construction, provide customized thinking development programs for students of different levels and needs, and offer teachers scientific support for teaching decisions, thereby contributing to the overall improvement of the quality of education and teaching. Summary of the Invention

[0005] To address the aforementioned problems in the existing technology, this invention provides a personalized thinking training system driven by intelligent questioning; The objective of this invention can be achieved through the following technical solutions: The module includes a data acquisition module, a guidance strategy modeling module, a dynamic progressive questioning interaction module, and a learning analysis and optimization suggestion module. The data acquisition module is used to receive teacher input parameters, load guidance rules and case libraries that match student backgrounds, and obtain guidance data. The guidance strategy modeling module performs adaptability verification on the guidance data, obtains the questioning style based on the verification results and student background; it also performs vocabulary decomposition based on teaching objectives, establishes a mapping relationship between the guidance data, the teaching objectives and the guidance strategy, and obtains guidance configuration data. The dynamic progressive questioning interaction module, based on the guidance configuration data and combined with the student's input viewpoints, uses the questioning submodule to generate follow-up questions on the transfer of the teaching objectives; at the same time, it detects the student's status and triggers branch processing, obtains progressive questioning data and interaction status data, and stores the questioning data and the interaction status data; The learning analysis and optimization suggestion module is used to analyze the stored data and generate a structured evaluation report that guides optimization suggestions.

[0006] As a preferred technical solution of the present invention, the specific process of loading the guidance rules and case library that match the student background includes: extracting student background features and obtaining a student profile tag set; matching a subset of rules that are suitable for the thinking guidance logic from the guidance rule library based on the student profile tag set; filtering a subset of cases that correspond to the student profile tag set according to the preset question standards of the case library; and integrating the matched subset of rules with the subset of cases to obtain the resource components of the guidance data.

[0007] Specifically, the process of performing the adaptability verification includes: performing an integrity scan on the resource components in the guidance data, identifying missing parameter items and generating supplementary prompts; and using a semantic matching algorithm to verify the relevance between the teaching objectives and the theme, and the adaptability between the student's background and the guidance approach.

[0008] Specifically, the vocabulary decomposition based on teaching objectives includes: using a semantic parsing model to analyze the teaching objectives and extract conceptual vocabulary; constructing a vocabulary graph based on the conceptual vocabulary, and combining the guidance data to select semantic words that are suitable for students' cognitive levels; integrating the vocabulary graph with the selection results to obtain the semantic features of the vocabulary decomposed using teaching objectives.

[0009] Specifically, the mapping relationship is as follows: based on the vocabulary map and the student profile tags, a four-dimensional mapping matrix is ​​established and assigned corresponding weights through the logic of student profiles, vocabulary features, guidance rules, and cases; the weights of the mapping matrix are optimized based on historical guidance data to obtain a structured mapping table.

[0010] Specifically, the process of optimizing the mapping matrix weights based on historical guidance data includes: using a guidance effect attribution algorithm to extract student profiles, guidance strategies, and target achievement rates from historical guidance data, and obtaining the adaptation contribution value of the guidance strategy to students; constructing a dynamic weight adjustment model based on the contribution value, using target achievement rate, student participation, and opinion correction efficiency as weight factors, and quantifying the effect coefficient of the weight factors on the mapping relationship; using an incremental learning update mechanism to extract features from newly added guidance data in real time, and synchronously iteratively optimizing the parameters of the dynamic weight adjustment model; obtaining the updated four-dimensional mapping matrix, generating guidance configuration data, and providing dynamic weight support.

[0011] Specifically, the process of generating transfer questions for teaching objectives includes: based on the structured mapping table and the lexical graph, using the sequential execution logic of the progressive questioning submodule, generating transfer questions for teaching objectives and verifying the transfer questions.

[0012] Specifically, the sequential execution logic of the progressive questioning submodule includes: using a questioning node trigger threshold mechanism to set trigger condition thresholds for the four types of submodules: definition clarification, hypothesis testing, logical reductio ad absurdum, and summary transfer; obtaining a semantic anchoring mechanism through the submodules to semantically bind the follow-up question feedback results of the first submodule with the teaching target vocabulary to generate the input constraints of the second submodule; dynamically adjusting the follow-up question logic and question level of the next submodule based on the quality of the student's response to the previous submodule; and verifying the fit between the output of the fourth submodule and the teaching objectives through closed-loop verification logic.

[0013] Specifically, the process of detecting student status includes: collecting three types of status data: student response duration, response text, and logical expression of viewpoints; comparing the hierarchical graph of student viewpoints with the vocabulary using a semantic similarity algorithm to obtain a judgment result; and using a logical consistency detection mechanism to analyze the compliance of the judgment result and obtain the analysis result of student questioning status.

[0014] Specifically, the branching process is as follows: based on the analysis results of the student's questioning state, three types are identified: stuck state, off-topic state, and logical conflict state; for the stuck state, based on the case resources in the guidance configuration data, sub-questions and basic cases of preset difficulty are loaded; for the off-topic state, related follow-up questions are generated to bring the topic back; for the logical conflict state, based on the case resources, reverse cases are matched and contradictions are broken down to generate reductio ad absurdum follow-up questions.

[0015] Specifically, the analysis of the stored data includes: extracting three types of data based on the progressive questioning data and interaction status data: student response text, follow-up question response efficiency, and viewpoint correction trajectory; analyzing logical consistency based on the viewpoint correction trajectory, detecting the number of logical loopholes in the student response text, and statistically analyzing response speed, response length, and number of proactive questions; calculating participation scores based on the follow-up question response efficiency; and analyzing the degree of alignment between the number of corrections, correction direction, and teaching objectives to obtain performance indicators of the stored data.

[0016] Specifically, the generation of guidance optimization suggestions includes: identifying students' ability weaknesses based on the performance indicators and the results of the teaching objective achievement calculation; matching guidance rules and case resources that are suitable for improving weaknesses based on the mapping relationship table; optimizing the trigger threshold and processing logic; and generating guidance optimization suggestions.

[0017] The beneficial effects of this invention are: stimulating students' active thinking and deepening the effect of thinking training. Abandoning didactic questioning, the program guides students to independently organize their logic through dynamic and progressive questioning, strengthening their initiative in thinking. The transfer questioning and branching processing mechanisms focus on teaching objectives, helping students focus on core issues and improve their logical reasoning and problem-solving abilities.

[0018] Based on student profile matching rules and cases, the difficulty of questioning is adjusted according to cognitive level to avoid a one-size-fits-all approach. The dynamic weight adjustment model continuously optimizes the guidance strategy, providing adaptive training programs for students with different foundations and interests; vocabulary decomposition and a four-dimensional mapping matrix construct a structured guidance path to avoid fragmented follow-up questions that deviate from the teaching objectives.

[0019] A closed-loop verification logic and semantic anchoring mechanism ensure that questioning always revolves around the core of teaching, improving the goal achievement rate. Real-time monitoring of student status promptly identifies logical flaws and thinking deviations, allowing for targeted correction. The system generates structured reports containing performance indicators and optimization suggestions, providing teachers with quantitative decision-making support and driving iterative improvement in guidance quality. Attached Figure Description

[0020] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0021] Figure 1 This is a flowchart illustrating a personalized thinking training system driven by intelligent questioning, as described in this invention. Figure 2 This is a flowchart of the dynamic progressive questioning interaction in this invention. Detailed Implementation

[0022] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0023] Please see Figure 1-2 A personalized thinking training system driven by intelligent questioning, comprising: The module includes a data acquisition module, a guidance strategy modeling module, a dynamic progressive questioning interaction module, and a learning analysis and optimization suggestion module. The data acquisition module is used to receive teacher input parameters, load guidance rules and case libraries that match student backgrounds, and obtain guidance data. The guidance strategy modeling module performs adaptability verification on the guidance data, obtains the questioning style based on the verification results and student background; it also performs vocabulary decomposition based on teaching objectives, establishes a mapping relationship between the guidance data, the teaching objectives and the guidance strategy, and obtains guidance configuration data. The dynamic progressive questioning interaction module, based on the guidance configuration data and combined with the student's input viewpoints, uses the questioning submodule to generate follow-up questions on the transfer of the teaching objectives; at the same time, it detects the student's status and triggers branch processing, obtains progressive questioning data and interaction status data, and stores the questioning data and the interaction status data; The learning analysis and optimization suggestion module is used to analyze the stored data and generate a structured evaluation report that guides optimization suggestions.

[0024] As a preferred technical solution of the present invention, the specific process of loading the guidance rules and case library that match the student background includes: extracting student background features and obtaining a student profile tag set; matching a subset of rules that are suitable for the thinking guidance logic from the guidance rule library based on the student profile tag set; filtering a subset of cases that correspond to the student profile tag set according to the preset question standards of the case library; and integrating the matched subset of rules with the subset of cases to obtain the resource components of the guidance data.

[0025] In this embodiment, teachers input core parameters to clarify teaching objectives, training duration range, difficulty range, and core test points, providing the system with basic training directions and ensuring that training content is precisely aligned with teaching needs.

[0026] By extracting student background characteristics and comprehensively acquiring key information such as students' knowledge base, logical analysis ability, past learning weaknesses, interests, and learning habits through system-linked academic records, prior training data, and classroom performance records, a complete student profile tag set is formed to accurately depict individual student characteristics. Based on this student profile tag set, a subset of rules suitable for corresponding thinking guidance logic is selected from the guidance rule library. Differentiated guidance logics are matched for students with different knowledge bases and ability levels. For example, guidance rules focusing on consolidating basic concepts are suitable for students with weak foundations, while guidance rules focusing on in-depth exploration and expansion are suitable for students with strong abilities, ensuring that the guidance logic is highly matched with students' abilities and needs.

[0027] Based on the pre-defined criteria of the case library, such as question standards, difficulty levels, and scenario types, a subset of cases is selected that matches the student profile tag set. The difficulty, type, and presentation format of the cases are all aligned with students' cognitive levels and interests. For example, for students who prefer everyday scenarios, cases that are close to daily life are prioritized; for students who excel at abstract reasoning, cases with a certain degree of critical thinking are selected to enhance student participation. The matched rule subset and case subset are systematically integrated, supplemented with necessary auxiliary explanatory information and guiding logic, forming the core resource components of the guidance data, providing comprehensive and suitable basic data support for subsequent guidance strategy modeling.

[0028] Specifically, the process of performing the compatibility verification includes: performing an integrity scan on the resource components in the guidance data, identifying missing parameter items and generating supplementary prompts; and using a semantic matching mechanism to verify the relevance between the teaching objectives and the theme, and the compatibility between the student's background and the guidance approach.

[0029] In this embodiment, a comprehensive and detailed inspection of the resource components in the guidance data is conducted. Key elements such as core parameters, case materials, and guidance logic are verified one by one. Missing key parameters, logical gaps, or material vulnerabilities are identified, and clear and specific supplementary prompts are automatically generated. After the relevant parameters are supplemented and vulnerabilities are fixed, the integrity verification is completed. The semantic adaptation enhancement matching algorithm, with the following formula, deeply analyzes the semantic relevance between teaching objectives and case themes to ensure that the case content closely adheres to the core teaching objectives. Simultaneously, it verifies the compatibility between students' backgrounds and guidance approaches, judging whether the guidance methods and pace match students' learning habits and ability boundaries, ensuring that the guidance data fully meets training needs. Based on the compatibility verification results and student background characteristics, differentiated questioning styles are set for different students. For introverted students with weak foundations, a gentle and inspiring questioning style is adopted, with friendly language, a relaxed pace, and more encouraging guidance. For outgoing students with strong abilities, an interactive and exploratory questioning style is adopted, with concise and sharp language, focusing on stimulating critical thinking. This ensures that the questioning methods align with students' learning habits and improves the guidance effect.

[0030] Specifically, the vocabulary decomposition based on teaching objectives includes: using a semantic parsing model to analyze the teaching objectives and extract conceptual vocabulary; constructing a vocabulary graph based on the conceptual vocabulary, and combining the guidance data to select semantic words that are suitable for students' cognitive levels; integrating the vocabulary graph with the selection results to obtain the semantic features of the vocabulary decomposed using teaching objectives.

[0031] In this embodiment, the teaching objective vocabulary semantic deconstruction algorithm is used in the vocabulary decomposition process of teaching objectives. Through semantic parsing and graph construction, it accurately extracts vocabulary semantic features that are suitable for students' cognitive levels, providing a foundation for the subsequent establishment of mapping relationships. Its formula is: , F 语义 The integrated lexical semantic feature vector; n is the number of semantic words that are selected and adapted to the students' cognitive level; V represents the importance weight of the i-th semantic word (set based on the relevance to teaching objectives, with a value range of [0.1, 1.0]); i M is the word vector of the i-th semantic word (generated through a semantic parsing model); i,j This is the association strength matrix between the i-th semantic word and the j-th core concept word (value range [0,1]).

[0032] Specifically, the mapping relationship is as follows: based on the vocabulary map and the student profile tags, a four-dimensional mapping matrix is ​​established and assigned corresponding weights through the logic of student profiles, vocabulary features, guidance rules, and cases; the weights of the mapping matrix are optimized based on historical guidance data to obtain a structured mapping table.

[0033] Specifically, the process of optimizing the mapping matrix weights based on historical guidance data includes: using a guidance effect attribution algorithm to extract student profiles, guidance strategies, and target achievement rates from historical guidance data, and obtaining the adaptation contribution value of the guidance strategy to students; constructing a dynamic weight adjustment model based on the contribution value, using target achievement rate, student participation, and opinion correction efficiency as weight factors, and quantifying the effect coefficient of the weight factors on the mapping relationship; using an incremental learning update mechanism to extract features from newly added guidance data in real time, and synchronously iteratively optimizing the parameters of the dynamic weight adjustment model; obtaining the updated four-dimensional mapping matrix, generating guidance configuration data, and providing dynamic weight support.

[0034] In this embodiment, based on the set of lexical semantic features and the set of student profile tags, a four-dimensional mapping matrix is ​​established with student profile, lexical features, guidance rules, and cases as the four core dimensions. Combined with teaching experience and initial data, reasonable initial weights are assigned to each dimension to build a basic mapping framework.

[0035] A guidance effect attribution algorithm is employed to extract relevant samples of student profiles, guidance strategies, and goal achievement from historical guidance data. The algorithm analyzes the suitability of different guidance strategies for different types of students, obtaining the contribution value of each strategy to student suitability. Based on this contribution value, a dynamic weight adjustment model is constructed, using key indicators such as goal achievement rate, student participation, and opinion correction efficiency as core weight factors. The algorithm quantifies the coefficient of each factor's influence on the mapping relationship. An incremental learning update mechanism is used to extract and analyze new guidance data in real time, iteratively optimizing the parameters of the dynamic weight adjustment model to ensure the four-dimensional mapping matrix continuously reflects the latest training data. Finally, an updated four-dimensional mapping matrix is ​​obtained, generating guidance configuration data supported by dynamic weights.

[0036] Specifically, the process of generating transfer questions for teaching objectives includes: based on the structured mapping table and the lexical graph, using the sequential execution logic of the progressive questioning submodule, generating transfer questions for teaching objectives and verifying the transfer questions.

[0037] Specifically, the sequential execution logic of the progressive questioning submodule includes: using a questioning node trigger threshold mechanism to set trigger condition thresholds for the four types of submodules: definition clarification, hypothesis testing, logical reductio ad absurdum, and summary transfer; obtaining a semantic anchoring mechanism through the submodules to semantically bind the follow-up question feedback results of the first submodule with the teaching target vocabulary to generate the input constraints of the second submodule; dynamically adjusting the follow-up question logic and question level of the next submodule based on the quality of the student's response to the previous submodule; and verifying the fit between the output of the fourth submodule and the teaching objectives through closed-loop verification logic.

[0038] In this embodiment, the threshold mechanism adopts a progressive interrogation-based dynamic triggering threshold algorithm, the formula of which is: , T k The dynamic trigger threshold for the k-th type of interrogation submodule (k=1,2,3,4 correspond to the four types of submodules respectively); T k0 Let be the initial trigger threshold for the k-th submodule; Q represents the quality score of the student's response to the previous submodule (calculated based on semantic completeness, logical accuracy, and relevance). In response to the average quality benchmark; This is the sensitivity coefficient, which controls the threshold adjustment range.

[0039] Dynamic trigger thresholds are set for four types of interrogation sub-modules: definition clarification, hypothesis testing, logical absurdity, and summary transfer. These thresholds are adjusted based on the quality of students' real-time responses to achieve precise connection and difficulty adaptation of the interrogation logic.

[0040] The Definition Clarification submodule addresses core concepts and easily confused knowledge points in the teaching objectives by posing precise clarification questions to students, helping them understand key definitions, connotations, and applicable boundaries, thus solidifying their knowledge foundation. The Hypothesis Testing submodule, based on students' responses to the definition clarification questions and combined with the logical connections in the vocabulary map, poses reasonable hypothetical questions, guiding students to connect conceptual knowledge with specific scenarios to verify the accuracy and comprehensiveness of their understanding. The Logical Reduction Submodule addresses potential logical loopholes and thinking biases in students' responses by designing targeted reduction-of-fact questions. Through progressively deeper questioning, it encourages students to independently discover their own thinking problems and clarify logical relationships. The Summary and Transfer submodule guides students to systematically summarize core knowledge points and thinking methods, extracting patterns, and poses cross-scenario transfer questions to help students apply their learned knowledge and methods to new, similar, or extended scenarios, achieving the transfer and improvement of knowledge and thinking abilities.

[0041] Through closed-loop verification logic, the alignment between the output of the transfer submodule and the teaching objectives is continuously verified and summarized. If any deviation from the core teaching is found, the direction of follow-up questions is adjusted in a timely manner to ensure that the follow-up questions always revolve around the core teaching.

[0042] Specifically, the process of detecting student status includes: collecting three types of status data: student response duration, response text, and logical expression of viewpoints; comparing the hierarchical graph of student viewpoints with the vocabulary using a semantic similarity algorithm to obtain a judgment result; and using a logical consistency detection mechanism to analyze the compliance of the judgment result and obtain the analysis result of student questioning status.

[0043] In this embodiment, during the interaction process, multi-dimensional state data such as the student's response duration, response text content, logical expression of opinions, and emotional feedback (such as whether there are frequent pauses or hesitant tone) are collected in real time and comprehensively to construct a panoramic view of the student's interaction state.

[0044] By using semantic similarity algorithms, students' viewpoints are compared with the vocabulary hierarchy map to accurately obtain the degree of fit and deviation between students' viewpoints and core knowledge points, thus clarifying students' mastery of knowledge. A logical consistency detection mechanism is used to analyze the compliance of the judgment results and to deeply identify whether students have thinking block, logical confusion, deviation from the topic, or contradictory viewpoints, forming a comprehensive and accurate analysis of students' questioning status.

[0045] Specifically, the branching process is as follows: based on the analysis results of the student's questioning state, three types are identified: stuck state, off-topic state, and logical conflict state; for the stuck state, based on the case resources in the guidance configuration data, sub-questions and basic cases of preset difficulty are loaded; for the off-topic state, related follow-up questions are generated to bring the topic back; for the logical conflict state, based on the case resources, reverse cases are matched and contradictions are broken down to generate reductio ad absurdum follow-up questions.

[0046] In this embodiment, state type identification: based on the analysis results of the student's questioning state, accurately identify three common abnormal state types: stuck state, off-topic state, and logical conflict state, and clarify the essence of the problem.

[0047] Stuck State: When students experience mental stagnation, are unable to respond for a long time, or have broken logic in their responses, the system loads sub-problems of preset difficulty based on the case resources in the guidance configuration data. This breaks down complex problems into simple and easy-to-understand step-by-step problems, and provides basic case studies to aid understanding, reducing the difficulty of thinking, helping students break through mental bottlenecks, and gradually return to the normal interaction rhythm.

[0048] Off-topic state: When students' responses deviate from the teaching objectives and core knowledge points, generate follow-up questions that are highly relevant to the teaching objectives and connect with the students' current viewpoints. In a natural and gentle way, bring students' thinking back and guide them to refocus on the core issues and start thinking.

[0049] Logical Conflict Status: When a student's viewpoint contains internal logical contradictions or conflicts with core knowledge points, targeted reverse cases are matched based on the case library. The contradictions in the student's viewpoint are broken down layer by layer to generate precise reductio ad absurdum questions. This helps students intuitively recognize their own logical problems, clarify logical relationships, and correct thinking biases.

[0050] Specifically, the analysis of the stored data includes: extracting three types of data based on the progressive questioning data and interaction status data: student response text, follow-up question response efficiency, and viewpoint correction trajectory; analyzing logical consistency based on the viewpoint correction trajectory, detecting the number of logical loopholes in the student response text, and statistically analyzing response speed, response length, and number of proactive questions; calculating participation scores based on the follow-up question response efficiency; and analyzing the degree of alignment between the number of corrections, correction direction, and teaching objectives to obtain performance indicators of the stored data.

[0051] Specifically, the generation of guidance optimization suggestions includes: identifying students' ability weaknesses based on the performance indicators and the results of the teaching objective achievement calculation; matching guidance rules and case resources that are suitable for improving weaknesses based on the mapping relationship table; optimizing the trigger threshold and processing logic; and generating guidance optimization suggestions.

[0052] In this embodiment, based on the performance indicators and the results of the teaching objective achievement calculation, the system accurately identifies students' weaknesses in knowledge mastery (such as incomplete understanding of concepts, omissions of knowledge points, etc.), logical analysis (such as sloppy logical reasoning, lack of systematicity, etc.), and thinking application (such as weak knowledge transfer ability, insufficient innovative thinking, etc.), thus clarifying the key areas for subsequent improvement. Based on the mapping relationship table, the system specifically matches guidance rules and case resources to improve students' ability weaknesses, optimizes the questioning trigger threshold and processing logic, and generates personalized and actionable guidance optimization suggestions. For example, for students with weak logical analysis ability, guidance rules and cases focusing on logical decomposition are recommended; for students with insufficient knowledge transfer ability, guidance content for cross-scenario transfer training is added. The system integrates student performance indicator radar charts, in-depth analysis of ability weaknesses, personalized guidance optimization suggestions, and references for subsequent training plans to generate a well-organized and focused structured evaluation report. This provides teachers with a scientific and quantitative basis for adjusting teaching strategies and conducting precise teaching, and also allows for feedback on training progress to students and parents, helping home-school collaboration to improve training effectiveness.

[0053] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A personalized thinking training system driven by intelligent questioning, characterized in that, include: The module includes a data acquisition module, a guidance strategy modeling module, a dynamic progressive questioning interaction module, and a learning analysis and optimization suggestion module. The data acquisition module is used to receive teacher input parameters, load guidance rules and case libraries that match student backgrounds, and obtain guidance data. The guidance strategy modeling module performs adaptability verification on the guidance data, obtains the questioning style based on the verification results and student background; it also performs vocabulary decomposition based on teaching objectives, establishes a mapping relationship between the guidance data, the teaching objectives and the guidance strategy, and obtains guidance configuration data. The dynamic progressive questioning interaction module, based on the guidance configuration data and combined with the student's input viewpoints, uses the questioning submodule to generate follow-up questions on the transfer of the teaching objectives; at the same time, it detects the student's status and triggers branch processing, obtains progressive questioning data and interaction status data, and stores the questioning data and the interaction status data; The learning analysis and optimization suggestion module is used to analyze the stored data and generate a structured evaluation report that guides optimization suggestions.

2. The system according to claim 1, characterized in that, The specific process of loading the guidance rules and case library that match the student's background includes: extracting student background features and obtaining a student profile tag set; matching a subset of rules that fit the thinking guidance logic from the guidance rule library based on the student profile tag set; filtering a subset of cases that correspond to the student profile tag set according to the preset question standards of the case library; and integrating the matched subset of rules with the subset of cases to obtain the resource components of the guidance data.

3. The system according to claim 1, characterized in that, The specific process of performing the adaptability verification includes: performing an integrity scan on the resource components in the guidance data, identifying missing parameter items and generating supplementary prompts; and using a semantic matching algorithm to verify the relevance between the teaching objectives and the theme, and the adaptability between the student's background and the guidance approach.

4. The system according to claim 1, characterized in that, The vocabulary decomposition based on teaching objectives includes: using a semantic parsing model to parse the teaching objectives and extract conceptual vocabulary; constructing a vocabulary graph based on the conceptual vocabulary, and combining the guidance data to select semantic words that are suitable for students' cognitive levels; integrating the vocabulary graph with the selection results to obtain the semantic features of the vocabulary decomposed using teaching objectives.

5. The system according to claim 1, characterized in that, The mapping relationship is as follows: based on the vocabulary map and the student profile tags, a four-dimensional mapping matrix is ​​established and assigned corresponding weights through the logic of student profiles, vocabulary features, guidance rules, and cases; the weights of the mapping matrix are optimized based on historical guidance data to obtain a structured mapping table.

6. The system according to claim 1, characterized in that, The specific process of optimizing the mapping matrix weights based on historical guidance data includes: using a guidance effect attribution algorithm to extract student profiles, guidance strategies, and target achievement rates from historical guidance data, and obtaining the adaptation contribution value of the guidance strategy to students; constructing a dynamic weight adjustment model based on the contribution value, using target achievement rate, student participation, and opinion correction efficiency as weight factors, and quantifying the effect coefficient of the weight factors on the mapping relationship; using an incremental learning update mechanism to extract features from newly added guidance data in real time, and synchronously iteratively optimizing the parameters of the dynamic weight adjustment model; obtaining the updated four-dimensional mapping matrix, generating guidance configuration data, and providing dynamic weight support.

7. The system according to claim 1, characterized in that, The specific process of generating transfer questions for teaching objectives includes: based on the structured mapping table and the lexical graph, using the sequential execution logic of the progressive interrogation submodule, generating transfer questions for teaching objectives and verifying the transfer questions.

8. The system according to claim 7, characterized in that, The specific process of the sequential execution logic of the progressive questioning submodule includes: adopting a questioning node trigger threshold mechanism to set trigger condition thresholds for the four types of submodules: definition clarification, hypothesis testing, logical reductio ad absurdum, and summary transfer; obtaining a semantic anchoring mechanism through the submodules to semantically bind the follow-up question feedback results of the first submodule with the teaching target vocabulary to generate the input constraints of the second submodule; dynamically adjusting the follow-up question logic and question level of the next submodule based on the quality of the student's response to the previous submodule; and verifying the fit between the output of the fourth submodule and the teaching objectives through closed-loop verification logic.

9. The system according to claim 1, characterized in that, The specific process for detecting student status includes: collecting three types of status data: student response duration, response text, and logical expression of viewpoints; comparing the hierarchical graph of student viewpoints with the vocabulary using a semantic similarity algorithm to obtain a judgment result; and using a logical consistency detection mechanism to analyze the compliance of the judgment result and obtain the analysis result of student questioning status.

10. The system according to claim 1, characterized in that, The branching process is as follows: based on the analysis results of the student's questioning state, three types are identified: stuck state, off-topic state, and logical conflict state; for the stuck state, based on the case resources in the guidance configuration data, sub-questions and basic cases of preset difficulty are loaded; for the off-topic state, related follow-up questions are generated to bring the topic back; for the logical conflict state, based on the case resources, reverse cases are matched and contradictions are broken down to generate reductio ad absurdum follow-up questions.

11. The system according to claim 1, characterized in that, The analysis of the stored data includes: extracting three types of data based on the progressive questioning data and interaction status data: student response text, follow-up question response efficiency, and viewpoint correction trajectory; analyzing logical consistency based on the viewpoint correction trajectory, detecting the number of logical loopholes in the student response text, and statistically analyzing response speed, response length, and number of proactive questions; calculating participation scores based on the follow-up question response efficiency; and analyzing the degree of alignment between the number of corrections, correction direction, and teaching objectives to obtain performance indicators of the stored data.

12. The system according to claim 1, characterized in that, The process of generating guidance optimization suggestions includes: identifying students' skill gaps based on the performance indicators and the results of the teaching objective achievement calculation; matching guidance rules and case resources that are appropriate for improving these gaps based on the mapping relationship table; optimizing trigger thresholds and processing logic; and generating guidance optimization suggestions.

Images