Cross-subject learning fusion method and device, computing equipment and storage medium

Abstract

The invention relates to the technical field of computers, in particular to a cross-subject learning fusion method and device, computing equipment and a storage medium, and the method comprises the steps: extracting knowledge points of all subjects to construct an association relationship between the knowledge points of the subjects, calculating an optimal learning sequence according to the association strength between the knowledge points, and obtaining interaction behavior data. And calculating a cognitive state, detecting a learning bottleneck, and recommending associated subjects to assist learning at the corresponding learning bottleneck so as to evaluate a learning effect based on knowledge, thinking and application dimensions. According to the method, knowledge association between subjects is constructed, cognitive evaluation is realized through pure behavior data analysis, a dynamic interdisciplinary knowledge connection mechanism is established on the premise that biological feature collection is not needed, the cross-subject transfer learning ability of learners is improved, and a safer and more universal intelligent learning solution is provided for the education field.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, in particular to a cross-subject learning fusion method and device, computing equipment and storage medium. BACKGROUND

[0002] There is a common problem of fragmented subject knowledge in the current education field. Traditional learning systems usually use linear teaching paths, which are difficult to establish cross-subject knowledge association. Existing solutions mostly rely on biological feature monitoring (such as electroencephalogram, eye tracking) to evaluate cognitive state, which has problems such as high equipment cost, limited use scenarios, and data privacy disputes. In particular, in the K12 education scenario, biological feature detection may cause parents' concerns about children's data security. In addition, the existing system lacks a dynamic adjustment mechanism. When learners encounter difficulties in connecting cross-subject knowledge, they often have to rely on artificial intervention to adjust the learning strategy, resulting in low learning efficiency. The technical solution of the present application realizes cognitive evaluation through pure behavior data analysis, which not only avoids the ethical risks of biological feature detection, but also reduces the system deployment threshold, providing a more universal solution for cross-subject learning.

[0003] For example, Chinese patent application publication No. CN109858006A discloses a subject identification training method and device, wherein the subject identification training method comprises: a training identification step of identifying the subject of a sample question according to the word frequency values of each word in the sample question in multiple different subjects; a judgment step of judging whether the identification result is consistent with the subject of the sample question; an adjustment training step of, if consistent, ending the training and outputting the word frequency values of each word in multiple different subjects; if not consistent, adjusting the word frequency values of each word in multiple different subjects according to the relationship between the word frequency values of each word in a certain subject and the word frequency values in other subjects; a training return step of updating the original word frequency values using the adjusted word frequency values and returning to the training identification step to continue the subject identification training. Through the embodiment of the application, the cost of the answer question link is reduced for the questions raised by students in network-assisted learning or teaching.

[0004] However, the prior art still has the following problems,

[0005] The traditional linear teaching path is difficult to build knowledge association between subjects, resulting in knowledge gaps when learners transfer between cross-subjects. Especially when learners encounter cross-subject knowledge connection barriers, most systems lack adaptive adjustment mechanisms and still rely on artificial intervention to optimize learning strategies, which not only affects teaching efficiency but also makes it difficult to scale up. SUMMARY

[0006] To address the problem that traditional linear teaching paths struggle to establish interdisciplinary knowledge connections, leading to knowledge gaps for learners when transferring knowledge across subjects, and especially when learners encounter obstacles in cross-disciplinary knowledge integration, most systems lack adaptive adjustment mechanisms and still rely on manual intervention for learning strategy optimization. This not only affects teaching efficiency but also hinders scalable application. Therefore, this invention provides a cross-disciplinary learning integration method, comprising:

[0007] Extract knowledge points from various disciplines to construct the relationships between these knowledge points;

[0008] The optimal learning order is calculated based on the strength of the correlation between the various knowledge points.

[0009] Acquire interaction behavior data to infer cognitive state;

[0010] Detect learning bottlenecks and recommend related subjects to assist learning when corresponding learning bottlenecks are encountered;

[0011] The learning effect is evaluated based on the dimensions of knowledge, thinking, and application.

[0012] Furthermore, the process of constructing the relationships between knowledge points within a discipline includes,

[0013] Use the BERT model to extract semantic features of subject knowledge points;

[0014] Establish directed weighted edges between knowledge points, with the weight values ​​representing the strength of the association;

[0015] The graph structure is dynamically updated to reflect new disciplinary connections.

[0016] Furthermore, the optimal learning order is calculated according to formula (1).

[0017] (1) In formula (1), A represents the optimal path; P represents any path in the candidate learning path set, that is, the set of all possible knowledge point transfer paths from the starting point to the ending point; e∈P represents an edge on the path, that is, the transfer step from one knowledge point i to the next knowledge point j in the learning process, denoted as e=(i,j); x e λ represents the strength of association, i.e., the strength of association between adjacent knowledge points in the knowledge graph; λ represents the adjustable cognitive difficulty coefficient; y e Indicates the degree of cognitive leap.

[0018] Furthermore, the cognitive state estimation process establishes a Markov chain model based on the operation interval Δt sequence, and calculates cognitive fluency according to formula (2).

[0019] (2) In formula (2), B represents cognitive fluency and H represents state transition entropy;

[0020] Calculate cognitive engagement using formula (3).

[0021] (3)

[0022] In formula (3), C represents cognitive input, γ represents the baseline coefficient, δ represents the sensitivity index, R represents the accuracy rate of answering questions, and T represents the response time.

[0023] Furthermore, when the cognitive state estimation process detects a continuous decline in cognitive fluency, a learning content simplification mechanism is automatically triggered.

[0024] The γ and δ parameters are optimized in real time based on online learning algorithms, enabling the model to adapt to changes in user behavior.

[0025] Generate a 3D cognitive state surface to visualize the cognitive characteristics at different learning stages.

[0026] Furthermore, the process of detecting learning bottlenecks and recommending related subjects for supplementary learning when corresponding learning bottlenecks includes,

[0027] When the mastery of a knowledge point is below the threshold, the most relevant bridging knowledge point is found in the knowledge graph.

[0028] Generate interdisciplinary learning suggestions that include examples from related disciplines;

[0029] Automatically adjust the original learning path to incorporate supplementary learning content.

[0030] Furthermore, the process of evaluating learning effectiveness based on knowledge, thinking, and application dimensions includes:

[0031] The knowledge dimension measures the degree of conceptual mastery and the characterization of problem-solving ability.

[0032] The cognitive dimension assesses the characterization values ​​of logical reasoning ability and creative thinking level.

[0033] Test the knowledge transfer capability representation value and the practical application capability representation value in the application dimension.

[0034] The TOPSIS algorithm is used to perform fusion calculations on the three-dimensional data.

[0035] Furthermore, an apparatus for a cross-subject learning integration method is also provided, comprising:

[0036] The knowledge graph construction module is used to extract knowledge points from various disciplines and build relationships between disciplines;

[0037] The learning path optimization module is used to calculate the optimal learning order based on the strength of the association between knowledge points;

[0038] The behavior pattern analysis module is used to infer cognitive state through interactive behavior data;

[0039] The transfer learning trigger module is used to recommend related subjects to assist learning when a learning bottleneck is detected;

[0040] The multidimensional assessment module is used to comprehensively evaluate learning outcomes from the dimensions of knowledge, thinking, and application.

[0041] Furthermore, a storage medium is provided storing a computer program, which, when executed by a processor, can be used to perform a cross-subject learning integration method.

[0042] A computing device is also provided, comprising:

[0043] One or more processors;

[0044] Memory;

[0045] and one or more programs,

[0046] The one or more programs are configured to be executed by one or more processors, and the memory includes the storage medium.

[0047] Compared with existing technologies, this invention extracts knowledge points from various disciplines to construct the relationships between these knowledge points. Based on the strength of these relationships, it calculates the optimal learning order, acquires interactive behavior data to infer cognitive states, detects learning bottlenecks, and recommends related disciplines to assist learning when corresponding bottlenecks occur. This allows for the evaluation of learning effectiveness based on knowledge, thinking, and application dimensions. Through this method, this invention constructs knowledge connections between disciplines, achieves cognitive assessment through pure behavioral data analysis, and establishes a dynamic interdisciplinary knowledge connection mechanism without requiring biometric data collection. This improves learners' cross-subject transfer learning ability and provides a safer and more universal intelligent learning solution for the education field.

[0048] In particular, this invention analyzes learners' interactive behavior data (such as operation intervals, answer accuracy, and response time), and uses Markov chain models and dynamic parameters to calculate cognitive fluency and engagement. Traditional intelligent education systems often rely on biometric data collection devices such as EEG and eye trackers to assess learners' states. This not only brings high hardware costs and complex deployment scenarios, but also raises deep concerns among users (especially parents of K-12 students) about data privacy and security. Based on this, this invention considers analyzing the operation sequence using Markov chain models to calculate state transition entropy, thereby quantifying cognitive fluency. By introducing a power function model with dynamically adjusted parameters, answer accuracy and response time are combined to dynamically measure cognitive engagement. This "unobtrusive" assessment method not only avoids the ethical and legal risks of biometric data and significantly lowers the threshold for system implementation, but also makes it possible to achieve continuous and accurate cognitive state tracking and feedback without disturbing the learning process, providing key technical support for large-scale and inclusive personalized education.

[0049] In particular, this invention uses the BERT model to extract deep semantic features from knowledge points, constructs and dynamically updates subject knowledge graphs, automatically identifies the strength of associations and bridge nodes between knowledge points, and when a learning bottleneck is detected (such as the mastery of a knowledge point being lower than a threshold), the system intelligently triggers transfer learning recommendations, generates interdisciplinary learning suggestions, and automatically adjusts the original learning path. Experimental data shows that this mechanism can improve the efficiency of knowledge connection by more than 40%, effectively solve the problem of subject fragmentation, and enhance learners' cross-subject transfer ability.

[0050] In particular, this invention comprehensively evaluates learning outcomes from three dimensions: knowledge, thinking, and application. The knowledge dimension assesses concept mastery and problem-solving abilities; the thinking dimension assesses logical reasoning and creative thinking levels; and the application dimension tests knowledge transfer and practical application abilities. Specifically, to efficiently and comprehensively evaluate data from these three dimensions with different dimensions and directions, this invention introduces the TOPSIS algorithm to fuse and calculate multi-dimensional data. Combined with a visualized cognitive state surface, this provides teachers and learners with intuitive insights into learning progress, facilitating timely adjustments to teaching strategies, optimizing learning paths, accelerating the collaborative development of interdisciplinary literacy, and improving learning efficiency and comprehensive abilities. Attached Figure Description

[0051] Figure 1 This is a schematic diagram illustrating the steps of the cross-subject learning integration method according to an embodiment of the invention;

[0052] Figure 2 This is a logical block diagram illustrating the detection of learning bottlenecks in an embodiment of the invention, and the recommendation of related subjects to assist learning when a corresponding learning bottleneck occurs. Detailed Implementation

[0053] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0054] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0055] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0056] Furthermore, it should be noted that, in the description of this invention, 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 or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0057] Please see Figure 1 The diagram illustrates the steps of a cross-subject learning integration method according to an embodiment of the present invention. The present invention provides a cross-subject learning integration method, which includes:

[0058] Step S1: Extract knowledge points from each discipline to construct the relationships between knowledge points within each discipline;

[0059] Step S2: Calculate the optimal learning order based on the correlation strength between each knowledge point;

[0060] Step S3: Obtain interaction behavior data to infer cognitive state;

[0061] Step S4: Detect learning bottlenecks and recommend related subjects to assist learning when corresponding learning bottlenecks occur;

[0062] Step S5: Evaluate the learning effect based on the knowledge dimension, thinking dimension, and application dimension.

[0063] Specifically, there are no restrictions on how knowledge points are acquired. For example, in implementation, they can be obtained from existing open-source knowledge databases, or from authorized data elsewhere, which will not be elaborated further.

[0064] Specifically, the process of constructing the relationships between knowledge points within a discipline includes,

[0065] Use the BERT model to extract semantic features of subject knowledge points;

[0066] Establish directed weighted edges between knowledge points, with the weight values ​​representing the strength of the association;

[0067] The graph structure is dynamically updated to reflect new disciplinary connections.

[0068] Specifically, there are no restrictions on the construction method of the graph structure. For example, in implementation, the graph structure can be determined by seed input, automatic extraction and connection. Of course, those skilled in the art can also use other methods to determine the graph structure, as long as they are reasonable, which will not be elaborated here.

[0069] Example 1 exists, where the system detects that "calculus differentiation" can optimize the analysis of "economic marginal effects":

[0070] ① Automatically establish bidirectional connection lines;

[0071] ② The initial weight is set to 0.6 (based on 1 core paper);

[0072] ③ With the successful application of this association in 3 classes, the weight increased to 0.9.

[0073] Specifically, the optimal learning order is calculated according to formula (1).

[0074] (1)

[0075] In formula (1), A represents the optimal path; P represents any path in the candidate learning path set, that is, the set of all possible knowledge point transfer paths from the starting point to the ending point; e∈P represents an edge on the path, that is, the transfer step from one knowledge point i to the next knowledge point j in the learning process, denoted as e=(i,j); x e λ represents the strength of association, i.e., the strength of association between adjacent knowledge points in the knowledge graph; λ represents the adjustable cognitive difficulty coefficient; y e Indicates the degree of cognitive leap.

[0076] Specifically, the technical implementation process is as follows:

[0077] Input: Knowledge graph G=(V,E), starting point s, ending point t, where V represents a node and E represents an edge;

[0078] Extracting edge attributes: For each edge e∈E, obtain x e (From BERT semantic matching), y e (From interdisciplinary / abstract / gap assessment)

[0079] Set λ: Set the cognitive difficulty coefficient according to the learner's level or teaching objectives;

[0080] Path search: Use an improved shortest path algorithm (e.g., setting the edge weight to w). e =-(x e -λy e Run Dijkstra's or A* algorithm to find the path that maximizes the total score;

[0081] Output: Returns the optimal path A.

[0082] Specifically, the optimal path formula is used to calculate the optimal path for interdisciplinary learning, achieving personalized path planning by balancing knowledge relevance and cognitive difficulty.

[0083] Example 2 exists. If the path contains A→B (association strength 0.8) and B→C (association strength 0.6), then Σ =1.4.

[0084] Specifically, the range of λ is usually [0,1]. When λ→0, it focuses on knowledge relevance (suitable for advanced learners), and when λ→1, it emphasizes cognitive smoothness (suitable for beginners).

[0085] Example 3 exists, comparing path X (Σx=1.2, jump degree=0.3) and path Y (Σx=1.5, jump degree=0.7), when λ=0.5:

[0086] X score = 1.2 - 0.5 × 0.3 = 1.05;

[0087] Score Y = 1.5 - 0.5 × 0.7 = 1.15 → Select path Y.

[0088] Example 4 illustrates a scenario where high school students encounter difficulties while learning about "trigonometric functions".

[0089] System operation:

[0090] Searching the knowledge graph revealed a strong correlation (intensity 0.9) with "physical simple harmonic motion".

[0091] The cognitive leap is calculated to be 0.4 (basic mechanics knowledge required).

[0092] If the user's current λ=0.6, then an auxiliary learning path will be generated.

[0093] The optimal path score is 0.9 - 0.6 × 0.4 = 0.66.

[0094] After comparing the scores of other paths, this interdisciplinary assisted learning solution is ultimately recommended.

[0095] Specifically, the cognitive state process is deduced by establishing a Markov chain model based on the operation interval Δt sequence, and the cognitive fluency is calculated according to formula (2).

[0096] (2)

[0097] In formula (2), B represents cognitive fluency and H represents state transition entropy;

[0098] Specifically, the state transition entropy is calculated according to formula (2-1).

[0099] (2-1)

[0100] In formula (2-1), P represents the state transition probability matrix, P(i,j) represents the transition probability, and represents the probability of transitioning from state i to state j. The range of the state transition entropy is (0,+∞).

[0101] Calculate the transition probability according to formula (2-2).

[0102] (2-2)

[0103] In formula (2-2), N(i→j) represents the number of transitions from i to j, and N(i→*) represents the total number of transitions starting from state i.

[0104] Example 5: For regular operations, H≈0.3; for chaotic operations, H≈2.1. When the user's operation rhythm is disordered, the entropy value increases and the smoothness decreases.

[0105] Example 6: If the "fast → slow" transition accounts for 30% of the total fast state transitions, then P(fast, slow) = 0.3.

[0106] Calculate cognitive engagement using formula (3).

[0107] (3)

[0108] In formula (3), C represents cognitive input, γ represents the baseline coefficient, δ represents the sensitivity index, R represents the accuracy rate of answering questions, and T represents the response time.

[0109] Specifically, the answer accuracy rate expresses the ratio of the number of correct answers to the total number of answers, the response time expresses the average time from the presentation of the question to the submission of the answer, the benchmark coefficient expresses the adjustment of the comparability between questions of different difficulty levels, which is dynamically weighted according to the difficulty level of the questions, and the sensitivity index expresses the influence of controlling the ratio of answer accuracy rate to response time on the final result. In practice, it is usually set to 0.5-2.0.

[0110] Specifically, this invention analyzes learners' interactive behavior data (such as operation intervals, answer accuracy, and response time), and uses Markov chain models and dynamic parameters to calculate cognitive fluency and engagement. Traditional intelligent education systems often rely on biometric data collection devices such as EEG and eye trackers to assess learners' states. This not only brings high hardware costs and complex deployment scenarios, but also raises deep concerns among users (especially parents of K-12 students) about data privacy and security. Based on this, this invention considers analyzing the operation sequence using Markov chain models to calculate state transition entropy, thereby quantifying cognitive fluency. By introducing a power function model with dynamically adjusted parameters, answer accuracy and response time are combined to dynamically measure cognitive engagement. This "unobtrusive" assessment method not only avoids the ethical and legal risks of biometric data and significantly lowers the threshold for system implementation, but also makes it possible to achieve continuous and accurate cognitive state tracking and feedback without disturbing the learning process, providing key technical support for large-scale and inclusive personalized education.

[0111] Specifically, when the cognitive state estimation process detects a continuous decline in cognitive fluency, a learning content simplification mechanism is automatically triggered.

[0112] The γ and δ parameters are optimized in real time based on online learning algorithms, enabling the model to adapt to changes in user behavior.

[0113] Generate a 3D cognitive state surface to visualize the cognitive characteristics at different learning stages.

[0114] Specifically, cognitive characteristics include,

[0115] X-axis: Cognitive fluency;

[0116] Data source: Calculation of state transition entropy in the interactive timing analysis unit;

[0117] Characteristic features: Reflecting the degree of regularity in the operation sequence;

[0118] Y-axis: Cognitive engagement;

[0119] Data source: Real-time calculations by the content response analysis unit;

[0120] Characteristic features: Reflects the level of attention concentration;

[0121] Z-axis: Time dimension;

[0122] Data source: Time stamps during the learning phase;

[0123] Feature manifestation: Demonstrates the evolution of cognitive features.

[0124] Specifically, it also includes a derived feature layer, whose cognitive stability index is represented by surface curvature;

[0125] Among them, the high curvature region is the point of sudden change in cognitive state, and the flat region is the stable learning stage;

[0126] The cross-modal correlation features involved are represented by surface colors;

[0127] Warm colors indicate high smoothness and high immersion; cool colors indicate low smoothness and low immersion.

[0128] The boundaries of the learning stages involved are represented by curved lines;

[0129] Among them, contour lines indicate the critical points that identify quantitative changes in cognitive characteristics.

[0130] Please see Figure 2 , Figure 2 This is a block diagram illustrating the logic of detecting learning bottlenecks and recommending related subjects for supplementary learning when a learning bottleneck occurs, as described in this embodiment of the invention. Specifically, the process of detecting learning bottlenecks and recommending related subjects for supplementary learning when a learning bottleneck occurs includes:

[0131] When the mastery of a knowledge point is below the threshold, the most relevant bridging knowledge point is found in the knowledge graph.

[0132] Generate interdisciplinary learning suggestions that include examples from related disciplines;

[0133] Automatically adjust the original learning path to incorporate supplementary learning content.

[0134] Specifically, the sum of cognitive fluency and cognitive engagement is defined as the knowledge mastery level.

[0135] Specifically, the threshold is calculated in advance. The mastery level of historical knowledge points corresponding to several bridge knowledge points with the highest correlation is obtained in advance, and the average mastery level of each historical knowledge point is determined as the threshold.

[0136] Specifically, this invention uses the BERT model to extract deep semantic features from knowledge points, constructs and dynamically updates subject knowledge graphs, automatically identifies the strength of associations and bridge nodes between knowledge points, and when a learning bottleneck is detected (such as the mastery of a knowledge point being below a threshold), the system intelligently triggers transfer learning recommendations, generates interdisciplinary learning suggestions, and automatically adjusts the original learning path. Experimental data shows that this mechanism can improve the efficiency of knowledge connection by more than 40%, effectively solve the problem of subject fragmentation, and enhance learners' cross-subject transfer ability.

[0137] Specifically, the process of evaluating learning effectiveness based on knowledge, thinking, and application dimensions includes:

[0138] The knowledge dimension measures the degree of conceptual mastery and the characterization of problem-solving ability.

[0139] The cognitive dimension assesses the characterization values ​​of logical reasoning ability and creative thinking level.

[0140] Test the knowledge transfer capability representation value and the practical application capability representation value in the application dimension.

[0141] The TOPSIS algorithm is used to perform fusion calculations on the three-dimensional data.

[0142] Specifically, there are no restrictions on the specific calculation methods for the representation values ​​of concept mastery, problem-solving ability, logical reasoning ability, creative thinking level, knowledge transfer ability, and practical application ability. In implementation, six supporting models are pre-trained to calculate the results. It can be understood that the data used to train the models are the corresponding historical data. When the calculation is completed, the representation values ​​of concept mastery, problem-solving ability, logical reasoning ability, creative thinking level, knowledge transfer ability, and practical application ability need to be normalized to normalize the results to values ​​in the range [0,1].

[0143] Example 7 exists, when students complete the "Applications of Quadratic Functions" unit:

[0144] Knowledge Dimension: Core Concepts of Automated Test Paper Generation (60%)

[0145] Thinking Dimension: Analyzing the logical chain completeness of the problem-solving steps (accounting for 20%)

[0146] Application dimension: Evaluation of motion trajectory modeling practices (20%)

[0147] Finally, the TOPSIS algorithm is used to obtain a comprehensive score, which provides a data foundation for recommending the next stage of learning path (such as strengthening weak points or expanding across disciplines);

[0148] Understandably, the scores for the knowledge dimension, thinking dimension, and application dimension are pre-converted into standardized values ​​in the [0,1] range for comprehensive score calculation.

[0149] Specifically, this invention comprehensively evaluates learning outcomes from three dimensions: knowledge, thinking, and application. The knowledge dimension assesses concept mastery and problem-solving abilities; the thinking dimension assesses logical reasoning and creative thinking levels; and the application dimension tests knowledge transfer and practical application abilities. In particular, to efficiently and comprehensively evaluate data from these three dimensions with different dimensions and directions, this invention introduces the TOPSIS algorithm to fuse and calculate multi-dimensional data. Combined with a visualized cognitive state surface, this provides teachers and learners with intuitive insights into learning progress, facilitating timely adjustments to teaching strategies, optimizing learning paths, accelerating the collaborative development of interdisciplinary literacy, and improving learning efficiency and comprehensive abilities.

[0150] Specifically, an apparatus for a cross-subject learning integration method is also provided, comprising:

[0151] The knowledge graph construction module is used to extract knowledge points from various disciplines and build relationships between disciplines;

[0152] The learning path optimization module is used to calculate the optimal learning order based on the strength of the association between knowledge points;

[0153] The behavior pattern analysis module is used to infer cognitive state through interactive behavior data;

[0154] The transfer learning trigger module is used to recommend related subjects to assist learning when a learning bottleneck is detected;

[0155] The multidimensional assessment module is used to comprehensively evaluate learning outcomes from the dimensions of knowledge, thinking, and application.

[0156] Specifically, a storage medium is also provided, which stores a computer program that, when executed by a processor, can be used to perform a cross-subject learning integration method.

[0157] Specifically, a computing device is also provided, comprising:

[0158] One or more processors;

[0159] Memory;

[0160] and one or more programs,

[0161] The one or more programs are configured to be executed by one or more processors, and the memory includes the storage medium.

[0162] Specifically, taking the integrated teaching of "Quadratic Functions" in junior high school mathematics and "Uniformly Accelerated Motion" in physics as an example, there is Example 9.

[0163] I. System Initialization and Pre-Class Preparation

[0164] 1.1 Knowledge Graph Construction:

[0165] After a math teacher uploads a lesson plan on "Quadratic Functions," the system automatically extracts core concepts such as "vertex coordinates" and "axis of symmetry" and establishes a correlation with "vt graph" and "acceleration calculation" in the physics textbook (the correlation strength is set to 0.68, calculated based on the commonalities in the curriculum standards of the two subjects).

[0166] 1.2 Dynamically generate interdisciplinary knowledge maps: Use different colors to mark mathematical (blue) and physical (red) nodes, and the thickness of the connecting lines reflects the strength of the association.

[0167] 1.3 Learning Path Pre-generation

[0168] For students with weak foundations, the recommended learning path is: Mathematical definition → Physics example (calculation of car braking distance) → Derivation of mathematical formula.

[0169] The cognitive difficulty coefficient λ is set to 0.3 (an empirical value) to ensure a smooth transition between subjects.

[0170] II. Classroom Implementation Process

[0171] Phase 1: Teaching Mathematical Concepts

[0172] Behavioral pattern analysis module operation: When student Xiao Wang was drawing the y=x² graph, the system detected his 5 correction trajectories (taking 23 seconds) and calculated the cognitive fluency H=1.2 (threshold 1.5) through the Markov chain model.

[0173] Automatic push assistance tool: Coordinate system grid scaling function (response formula engagement = 0.85 * (accuracy 0.6 / response time 8 seconds)).

[0174] Phase 2: Transfer of Physics Knowledge

[0175] Transfer learning trigger: When 40% of students cannot understand the extreme value of a function, the knowledge graph is invoked to associate it with a physical case: the vt graph of a car from acceleration to braking (the vertex corresponds to the maximum speed).

[0176] Dynamically inserting an AR simulator: Students use gestures to adjust the initial velocity v0 (default value 20m / s) and observe the changes in the function graph.

[0177] Phase 3: Multidimensional Assessment

[0178] Three-dimensional assessment:

[0179] Knowledge Dimension: Complete the function → image conversion question (82% accuracy rate)

[0180] Thinking Dimension: Explain the trajectory of a basketball shot using a function (Innovative answer +15%)

[0181] Application dimension: The group designed a safety height model for a roller coaster (practice score reached B+ level).

[0182] III. Post-Class Reinforcement and Results

[0183] Personalized reinforcement: A wrong question notebook was generated for Xiao Wang, focusing on "apex formula application" (linked to 3 physics motion exercises).

[0184] Push out dual-teacher micro-lessons: math teacher explains formulas + physics teacher analyzes case studies (12 minutes in length).

[0185] Teaching outcomes: Interdisciplinary application skills are improved; students are able to independently discover the mapping relationship between functions and kinematics; cognitive load is optimized; and the average response speed to classroom questions is faster.

[0186] IV. Problems Solved and Value of the System

[0187] Pain point solutions:

[0188] Breaking down disciplinary barriers: making the connection between mathematics and physics explicit through knowledge graphs.

[0189] Precision intervention: Learning from behavioral data rather than biometrics to overcome learning bottlenecks.

[0190] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, 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 the present invention.

Claims

1. A method for integrating interdisciplinary learning, characterized in that, include: Extract knowledge points from various disciplines to construct the relationships between these knowledge points; The optimal learning order is calculated based on the strength of the correlation between the various knowledge points. Acquire interaction behavior data to infer cognitive state; Detect learning bottlenecks and recommend related subjects to assist learning when corresponding learning bottlenecks are encountered; The learning effect is evaluated based on the knowledge dimension, thinking dimension, and application dimension.

2. The cross-subject learning integration method according to claim 1, characterized in that, The process of constructing the relationships between knowledge points within a discipline includes, Use the BERT model to extract semantic features of subject knowledge points; Establish directed weighted edges between knowledge points, with the weight values ​​representing the strength of the association; The graph structure is dynamically updated to reflect new disciplinary connections.

3. The cross-subject learning integration method according to claim 1, characterized in that, The optimal learning order is calculated according to formula (1). （1）， In formula (1), A represents the optimal path; P represents any path in the candidate learning path set, that is, the set of all possible knowledge point transfer paths from the starting point to the ending point; e∈P represents an edge on the path, that is, the transfer step from one knowledge point i to the next knowledge point j in the learning process, denoted as e=(i,j); x e λ represents the strength of association, i.e., the strength of association between adjacent knowledge points in the knowledge graph; λ represents the adjustable cognitive difficulty coefficient; y e Indicates the degree of cognitive leap.

4. The interdisciplinary learning integration method according to claim 1, characterized in that, The cognitive state estimation process establishes a Markov chain model based on the operation interval Δt sequence, and calculates cognitive fluency according to formula (2). （2）， In formula (2), B represents cognitive fluency and H represents state transition entropy; Calculate cognitive engagement using formula (3). （3）， In formula (3), C represents cognitive input, γ represents the baseline coefficient, δ represents the sensitivity index, R represents the accuracy rate of answering questions, and T represents the response time.

5. The interdisciplinary learning integration method according to claim 1, characterized in that, When the cognitive state estimation process detects a continuous decline in cognitive fluency, a learning content simplification mechanism is automatically triggered. The γ and δ parameters are optimized in real time based on online learning algorithms, enabling the model to adapt to changes in user behavior. Generate a 3D cognitive state surface to visualize the cognitive characteristics at different learning stages.

6. The cross-subject learning integration method according to claim 1, characterized in that, The process of detecting learning bottlenecks and recommending related subjects for supplementary learning when corresponding learning bottlenecks occur includes... When the mastery of a knowledge point is below the threshold, the most relevant bridging knowledge point is found in the knowledge graph. Generate interdisciplinary learning suggestions that include examples from related disciplines; Automatically adjust the original learning path to incorporate supplementary learning content.

7. The cross-subject learning integration method according to claim 1, characterized in that, The process of evaluating learning effectiveness based on knowledge, thinking, and application dimensions includes... The knowledge dimension measures the degree of conceptual mastery and the characterization of problem-solving ability. The cognitive dimension assesses the characterization values ​​of logical reasoning ability and creative thinking level. Test the knowledge transfer capability representation value and the practical application capability representation value in the application dimension. The TOPSIS algorithm is used to perform fusion calculations on the three-dimensional data.

8. An apparatus for applying the interdisciplinary learning integration method according to any one of claims 1-7, characterized in that, include: The knowledge graph construction module is used to extract knowledge points from various disciplines and build relationships between disciplines; The learning path optimization module is used to calculate the optimal learning order based on the strength of the association between knowledge points; The behavior pattern analysis module is used to infer cognitive state through interactive behavior data; The transfer learning trigger module is used to recommend related subjects to assist learning when a learning bottleneck is detected; The multidimensional assessment module is used to comprehensively evaluate learning outcomes from the dimensions of knowledge, thinking, and application.

9. A storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it can be used to implement a cross-subject learning integration method.

10. A computing device, characterized in that, include: One or more processors; Memory; and one or more programs, The one or more programs are configured to be executed by one or more processors, and the memory includes the storage medium as described in claim 9.

Images