Teaching big data analysis-based teacher-student matching degree evaluation method and system

Abstract

The invention relates to the technical field of teaching management, and discloses a teacher-student matching degree evaluation method and system based on teaching big data analysis, and the method comprises the steps: obtaining multi-source heterogeneous data generated by a teacher-student combination in a teaching process; performing cleaning and standardization processing on the multi-source heterogeneous data, and constructing a multi-dimensional feature vector of the teacher-student combination according to the processed multi-source heterogeneous data: inputting the multi-dimensional feature vector into a matching degree prediction model to obtain a specific quantized value of a multi-dimensional evaluation index of the teacher-student combination, and generating a dynamic weight value of the multi-dimensional evaluation index based on the historical matching effect data and the incremental teaching data so as to obtain a teacher-student matching degree evaluation result. The problems of single matching basis, low efficiency and the like are solved, personalized feature changes of students and teachers can be dynamically adapted, the matching accuracy and scientificity are remarkably improved, an efficient and intelligent teacher-student matching solution is provided for an online education platform, and personalized teaching and individualized material teaching are assisted.

Description

Technical Field

[0001] This invention relates to the field of teaching management technology, and in particular to a method and system for evaluating teacher-student matching based on teaching big data analysis. Background Technology

[0002] With the rapid development of information technology, online education platforms have sprung up like mushrooms after rain, and the number of teachers and students on these platforms continues to rise. The teaching process is essentially an interactive process between teachers and students. Different teachers have unique teaching styles and methods, and students also have individual differences, with varying degrees of adaptability to different teaching methods. Therefore, accurately matching students with teachers who suit their learning characteristics is crucial for improving teaching quality and student learning outcomes.

[0003] In traditional online teaching models, academic staff typically assign teachers who are proficient in a particular subject to students manually, based on their grade level and subject. However, this manual matching method has significant drawbacks: on the one hand, it consumes a large amount of manpower and time; on the other hand, the matching criteria are relatively singular, only considering the student's grade level and subject, ignoring the student's individual characteristics such as learning ability and learning style. This can lead to the matched teacher's teaching style not being suitable for the student's learning needs, thereby affecting the student's learning progress and growth.

[0004] While existing online education platforms offer teacher-student matching to some extent, most rely on simple rules or limited features, failing to fully utilize the massive amounts of data generated during the teaching process, such as student classroom performance, grades, and homework completion. This data contains rich information reflecting students' learning behaviors, habits, knowledge acquisition, and the effectiveness of teachers' instruction. In-depth mining and analysis of this teaching big data will provide a more accurate and scientific basis for teacher-student matching.

[0005] Furthermore, existing technologies often lack comprehensive and systematic evaluation indicators and methods for assessing teacher-student compatibility, making it difficult to accurately measure the degree of fit between teachers and students and hindering the realization of truly personalized teaching and individualized instruction. Therefore, how to fully utilize teaching big data to construct a scientific and reasonable teacher-student compatibility assessment system has become a key issue that urgently needs to be addressed in the current online education field. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention provides a method and system for evaluating teacher-student matching based on teaching big data analysis.

[0007] In a first aspect, the present invention provides a method for evaluating teacher-student matching based on teaching big data analysis, comprising the following technical solutions: Acquire multi-source heterogeneous data generated by the target teacher-student combination during the teaching process. The multi-source heterogeneous data includes: student classroom interaction behavior data, periodic assessment data, homework completion quality data, and teacher teaching video feature data, course knowledge point coverage data, and historical student evaluation data. The multi-source heterogeneous data is cleaned and standardized, and a multi-dimensional feature vector of the target teacher-student combination is constructed based on the processed multi-source heterogeneous data. The multidimensional feature vector is input into the matching degree prediction model to obtain the specific quantitative values ​​of the multidimensional evaluation indicators of the target teacher-student combination. Based on historical matching effect data and incremental teaching data, dynamic weight values ​​of the multidimensional evaluation indicators are generated to obtain the teacher-student matching degree evaluation results of the target teacher-student combination.

[0008] Furthermore, the multidimensional feature vector includes: a student multidimensional feature vector containing learning mode preference features, knowledge mastery features, and cognitive level features; and a teacher multidimensional feature vector containing teaching style features, knowledge transfer efficiency features, and interactive response ability features.

[0009] Furthermore, the step of constructing a multidimensional feature vector of the target teacher-student combination based on the processed multi-source heterogeneous data includes: The learning mode preference features are extracted from the classroom interaction behavior data, and the knowledge mastery features and cognitive level features are calculated based on the periodic assessment data and the homework completion quality data. The teaching style features are identified from the teaching video feature data, the knowledge transmission efficiency features are generated based on the course knowledge point coverage data, and the interactive response ability features are quantified through the historical student evaluation data.

[0010] Furthermore, the matching degree prediction model includes: a deep feature cross-network, a multi-head attention mechanism layer, a knowledge graph relationship reasoning module, a temporal behavior alignment model, a causal analysis model, and a multiple regression model; the multidimensional evaluation indicators include: knowledge transfer efficiency matching degree, learning style adaptation degree, interactive participation fit degree, and teaching feedback response degree.

[0011] Furthermore, the step of inputting the multidimensional feature vector into the matching degree prediction model to obtain the specific quantitative value of the multidimensional evaluation index of the target teacher-student combination includes: The deep feature cross-network is used to perform nonlinear interactive operations on the student multidimensional feature vector and the teacher multidimensional feature vector to generate a high-order cross-feature matrix. The higher-order cross-feature matrix is ​​input into the multi-head attention mechanism layer. By calculating the correlation scores between the feature dimensions of the higher-order cross-feature matrix, feature dimension weights are generated. The higher-order cross-feature matrix is ​​then weighted and fused to obtain an optimized teacher-student interaction representation. The optimized teacher-student interaction representation is input into the knowledge graph relationship reasoning module to match the teacher's knowledge point coverage path with the student's knowledge mastery structure, thereby obtaining a specific quantitative value of the knowledge transmission efficiency matching degree. Learning pattern preference features and teaching style features are extracted from the optimized teacher-student interaction representation, and combined with the classroom interaction behavior data and the teaching video feature data, the dynamic matching value is calculated through the temporal behavior alignment model to obtain the specific quantitative value of the learning style fit. Based on the interactive response ability characteristics and cognitive level characteristics in the optimized teacher-student interaction representation, the causal analysis model is used to correlate historical interaction records with changes in assessment scores to obtain a specific quantitative value of the interaction participation fit. By integrating the interactive response capability characteristics in the optimized teacher-student interaction representation with historical student evaluation data, the impact of feedback timeliness on knowledge consolidation rate is predicted through the multivariate regression model, thereby obtaining a specific quantitative value of the teaching feedback responsiveness.

[0012] Furthermore, the step of generating dynamic weight values ​​for the multidimensional evaluation indicators based on historical matching effect data and incremental teaching data includes: Based on the historical matching effect data, the initial weight values ​​of the multidimensional evaluation indicators are obtained, and the initial weight values ​​are calibrated online in conjunction with the teaching outcome feedback in the incremental teaching data to obtain the dynamic weight values.

[0013] Furthermore, the teacher-student matching degree assessment result is a teacher-student matching degree score; the step of obtaining the teacher-student matching degree assessment result of the target teacher-student combination includes: The teacher-student matching score of the target teacher-student combination is obtained by weighting and summing the specific quantitative values ​​and dynamic weight values ​​of the multidimensional evaluation indicators of the target teacher-student combination.

[0014] Furthermore, it also includes: Obtain teacher-student matching scores for multiple teacher-student pairs; A bipartite graph model for teacher-student matching is constructed, with students and teachers as nodes and teacher-student matching scores as edge weights. The minimum cost maximum flow algorithm is used to solve the bipartite graph model of teacher-student matching to obtain the globally optimal teacher-student pairing scheme.

[0015] Furthermore, the cleaning and standardization processes include at least one of the following: data interpolation completion, outlier filtering, normalization, noise removal, and structuring.

[0016] Secondly, this invention provides a teacher-student matching evaluation system based on teaching big data analysis, comprising the following technical solutions: The data acquisition module is used to acquire multi-source heterogeneous data generated by the target teacher-student combination during the teaching process. The multi-source heterogeneous data includes: student classroom interaction behavior data, periodic assessment data, homework completion quality data, and teacher teaching video feature data, course knowledge point coverage data, and historical student evaluation data. The processing module is used to clean and standardize the multi-source heterogeneous data, and construct a multi-dimensional feature vector of the target teacher-student combination based on the processed multi-source heterogeneous data. The evaluation module is used to input the multidimensional feature vector into the matching degree prediction model to obtain the specific quantitative values ​​of the multidimensional evaluation indicators of the target teacher-student combination, and to generate dynamic weight values ​​of the multidimensional evaluation indicators based on historical matching effect data and incremental teaching data, so as to obtain the teacher-student matching degree evaluation result of the target teacher-student combination.

[0017] This invention acquires multi-source heterogeneous data, cleans and standardizes it, constructs a multi-dimensional feature vector for the target teacher-student pairing, and combines it with a matching degree prediction model and dynamic weight values ​​to achieve accurate quantitative evaluation of teacher-student matching degree. This invention fully utilizes the massive amounts of data generated during the teaching process, overcomes the limitations of traditional manual matching, and solves problems such as single matching criteria and low efficiency. It can dynamically adapt to changes in the personalized characteristics of students and teachers, significantly improving the accuracy and scientific nature of matching. It provides online education platforms with an efficient and intelligent teacher-student matching solution, facilitating personalized teaching and individualized instruction.

[0018] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0020] Figure 1This is a flowchart illustrating a method for evaluating teacher-student matching based on teaching big data analysis. Figure 2 This is a schematic diagram of a teacher-student matching evaluation system based on teaching big data analysis. Detailed Implementation

[0021] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0022] Figure 1 This diagram illustrates a flowchart of an embodiment of a teacher-student matching evaluation method based on teaching big data analysis provided by the present invention. Figure 1 As shown, the method includes the following steps: S1. Obtain multi-source heterogeneous data generated by the target teacher-student combination during the teaching process. The multi-source heterogeneous data includes: student classroom interaction behavior data, periodic assessment data, homework completion quality data, as well as teacher teaching video feature data, course knowledge point coverage data, and historical student evaluation data.

[0023] The data includes: ① Classroom interaction behavior data, which represents students' real-time interactive behavior in the classroom, including: number of questions asked, correct answer rate, interaction frequency, and screen sharing duration. This data is used to analyze student participation and learning style preferences. ② Periodic assessment data, which represents the results of periodic learning outcome assessments, including: unit test scores, knowledge point mastery radar charts, and error distribution heatmaps. This data is used to quantify the dynamic changes in students' knowledge mastery and identify weaknesses. ③ Homework completion quality data, which represents quantitative indicators of students' performance on after-school tasks, including: on-time submission rate, homework pass rate, semantic completeness of subjective questions, and number of revision iterations. This data is used to assess students' learning attitudes and knowledge consolidation efficiency. ④ Teaching video feature data, which represents the results of multimodal analysis of teachers' lecture videos, including: speech rate fluctuation curves, blackboard writing density, animation demonstration frequency, and key paragraph replay rate. This data is used to identify teachers' teaching styles (e.g., lecture-style / guided style). ⑤ Course knowledge point coverage data refers to data representing the mapping relationship between course content and the subject knowledge system, including: knowledge point tree structure, core knowledge point markings, teaching progress association tables, etc. Course knowledge point coverage data is used to assess the fit between teachers' teaching content and the teaching syllabus. ⑥ Historical student evaluation data refers to data representing feedback information from previous students to teachers, including: quantitative scores, textual evaluations, and extraction of improvement suggestions, etc. Historical student evaluation data is used to quantify teachers' teaching improvement potential and interactive responsiveness.

[0024] It should be noted that the target teacher-student pairing is a randomly selected pair of teachers and students to be evaluated for teacher-student compatibility.

[0025] S2. The multi-source heterogeneous data is cleaned and standardized, and a multi-dimensional feature vector of the target teacher-student combination is constructed based on the processed multi-source heterogeneous data.

[0026] The multidimensional feature vectors include student and teacher multidimensional feature vectors. Student multidimensional feature vectors include learning mode preference features, knowledge mastery features, and cognitive level features. Teacher multidimensional feature vectors include teaching style features, knowledge transfer efficiency features, and interactive response ability features. Specifically: ① Learning mode preference features reflect students' adaptability to teaching methods, including: interaction tendency, information reception preference, and learning pace. ② Knowledge mastery features quantify students' understanding of specific knowledge modules, including: core concept mastery rate, error pattern classification, and stability assessment. ③ Cognitive level features measure the progression of students' thinking abilities, including: memory level, comprehension level, and application level. ④ Teaching style features describe typical teaching methods, including: lecture-style, guided, and demonstration-style. ⑤ Knowledge transfer efficiency features evaluate the matching efficiency between the teacher's teaching content and the target knowledge system, including: core knowledge point coverage, teaching path rationality, and depth adaptation. ⑥ Interactive response capability characteristics are used to quantify the quality of teachers' feedback to students' needs, including: response timeliness, feedback detail, and improvement implementation rate.

[0027] The cleaning and standardization methods include at least one of the following: data interpolation completion, outlier filtering, normalization, noise removal, and structuring.

[0028] In S2, the process of cleaning and standardizing multi-source heterogeneous data includes: ① using time series interpolation to complete missing classroom interaction behavior data; ② filtering outliers in the periodic assessment data using the interquartile range method; ③ normalizing the homework completion quality data using Min-Max standardization; ④ removing noisy segments in the teaching video feature data using spectral analysis; ⑤ structuring the course knowledge point coverage data using one-hot encoding; and ⑥ extracting sentiment scores from the text evaluations in the historical student evaluation data using the BERT model.

[0029] In S2, the steps for constructing a multidimensional feature vector of the target teacher-student combination based on the processed multi-source heterogeneous data include: The learning pattern preference features are extracted from the classroom interaction behavior data, and the knowledge mastery features and cognitive level features are calculated based on the periodic assessment data and the homework completion quality data.

[0030] Specifically: ① Behavioral indicators were constructed based on classroom interaction data, and clustering was performed using the K-means algorithm to obtain learning pattern preference features. ② Periodic assessment data and homework completion quality data were aligned, and scores for individual knowledge points were calculated separately to obtain the overall mastery level, i.e., the knowledge mastery feature. ③ Based on the periodic assessment data and homework completion quality data, the difficulty level of questions was graded, and scores for each level were calculated to obtain cognitive level features (memory level, comprehension level, and application level).

[0031] The teaching style features are identified from the teaching video feature data, the knowledge transmission efficiency features are generated based on the course knowledge point coverage data, and the interactive response ability features are quantified through the historical student evaluation data.

[0032] Specifically: ① The teaching video feature data is segmented and analyzed, and style indicators (lecturing tendency, guiding tendency, demonstration tendency, etc.) are calculated. The Analytic Hierarchy Process (AHP) is used to calculate the weight of each indicator. The type with the highest score after normalization is the teacher style label, i.e., the teaching style characteristic. ② Based on the course knowledge point coverage data, the matching degree between the teacher's actual explanation of knowledge points and the teaching syllabus is compared. A knowledge point dependency graph is constructed to detect whether the teacher's explanation order violates the precedence relationship, and the proportion of violations to the total number of knowledge points is calculated. Based on the above calculation results and the efficiency comprehensive scoring formula, the efficiency comprehensive score is obtained, i.e., the knowledge transfer efficiency characteristic. The efficiency comprehensive scoring formula is: ; This represents the overall efficiency score. This assesses the alignment between the knowledge points actually taught by teachers and the teaching syllabus. The percentage of violations out of the total number of knowledge points. ③ Use the BERT model to extract the sentiment polarity (-1~1) and improvement suggestion keywords (such as "slow grading" and "vague feedback") from historical student evaluation data, and combine them with predefined rules to obtain quantitative interactive response ability characteristics.

[0033] S3. Input the multidimensional feature vector into the matching degree prediction model to obtain the specific quantitative value of the multidimensional evaluation index of the target teacher-student combination, and generate the dynamic weight value of the multidimensional evaluation index based on historical matching effect data and incremental teaching data to obtain the teacher-student matching degree evaluation result of the target teacher-student combination.

[0034] The matching prediction model includes: a deep feature cross-network, a multi-head attention mechanism layer, a knowledge graph relationship reasoning module, a temporal behavior alignment model, a causal analysis model, and a multiple regression model. Multidimensional evaluation indicators include: knowledge transfer efficiency matching degree, learning style adaptability, interactive participation fit degree, and teaching feedback responsiveness. The teacher-student matching degree evaluation result is the teacher-student matching degree score.

[0035] In S3, the steps of inputting multidimensional feature vectors into the matching degree prediction model to obtain the specific quantitative values ​​of the multidimensional evaluation indicators of the target teacher-student combination include: The deep feature cross-network is used to perform nonlinear interactive operations on the student's multidimensional feature vector and the teacher's multidimensional feature vector to generate a high-order cross-feature matrix.

[0036] The deep feature cross-network includes a feature concatenation layer and a gated cross-network. Specifically, it integrates the student's multidimensional feature vectors. With teacher multidimensional feature vector Concatenate into a joint vector A gated cross network is used to... Feature interaction is performed, and three layers of cross-networks are stacked to generate a high-order cross-feature matrix. High-order cross feature matrix Each element in the expression represents the interaction strength between student features and teacher features.

[0037] The higher-order cross-feature matrix is ​​input into the multi-head attention mechanism layer. Feature dimension weights are generated by calculating the correlation scores between the feature dimensions of the higher-order cross-feature matrix. The higher-order cross-feature matrix is ​​then weighted and fused to obtain an optimized teacher-student interaction representation.

[0038] Among them, the multi-head attention mechanism layer obtains the optimized teacher-student interaction representation through multi-head attention calculation, weight fusion and normalization, and residual connection.

[0039] The optimized teacher-student interaction representation is input into the knowledge graph relationship reasoning module to match the teacher's knowledge point coverage path with the student's knowledge mastery structure, thereby obtaining a specific quantitative value of the knowledge transmission efficiency matching degree.

[0040] The knowledge graph relational reasoning module includes: constructing a knowledge graph, extracting knowledge point coverage features and mastery features from optimized representations, calculating the path matching degree, and introducing a penalty term to finally quantify the specific value of the knowledge transfer efficiency matching degree.

[0041] Learning pattern preference features and teaching style features are extracted from the optimized teacher-student interaction representation. Combined with the classroom interaction behavior data and the teaching video feature data, the dynamic matching value is calculated through the temporal behavior alignment model to obtain the specific quantitative value of the learning style fit.

[0042] Specifically, learning pattern preference features and teaching style features are extracted from optimized teacher-student interaction representations, and time-series sequences are constructed by combining them with original behavioral data: student behavior sequences. and teacher behavior sequence The student behavior sequence is calculated using the following formula. and teacher behavior sequence Minimum alignment cost: ; For the set of aligned paths; For the i-th action in the student action sequence, Let j be the j-th behavior in the teacher's behavior sequence. The learning style fit is quantified using the following formula to obtain a specific quantitative value: λ represents the attenuation coefficient (default value is 0.1). This represents a specific quantitative value indicating the fit of learning styles. For example, the peak student playback time is aligned with the teacher animation insertion time. ,but .

[0043] Based on the interactive response ability characteristics and cognitive level characteristics in the optimized teacher-student interaction representation, the causal analysis model is used to correlate historical interaction records with changes in assessment scores to obtain a specific quantitative value of the interaction participation fit.

[0044] Specifically, a causal directed acyclic graph is constructed, with interactive response capability characteristics as the cause and cognitive level hierarchy characteristics as the effect; Double machine learning is used to estimate the intervention effect and calculate the specific quantitative value of the degree of engagement in interactive participation. ; For interaction frequency, For the intervention effect, This is a specific quantitative value for the degree of engagement in interactive participation.

[0045] By integrating the interactive response capability characteristics in the optimized teacher-student interaction representation with historical student evaluation data, the impact of feedback timeliness on knowledge consolidation rate is predicted through the multivariate regression model, thereby obtaining a specific quantitative value of the teaching feedback responsiveness.

[0046] Specifically, the interactive response capability feature in the optimized teacher-student interaction representation is concatenated with historical student evaluation data; the Gradient Boosting Tree (GBRT) is used to predict the knowledge consolidation rate, and a specific quantitative value of the teaching feedback responsiveness is calculated: ; To predict knowledge retention rate, This represents the expected consolidation rate without feedback intervention. This is a specific quantitative value for the responsiveness of teaching feedback.

[0047] In S3, the steps for generating dynamic weight values ​​for multidimensional evaluation indicators based on historical matching effect data and incremental teaching data include: Based on the historical matching effect data, the initial weight values ​​of the multidimensional evaluation indicators are obtained, and the initial weight values ​​are calibrated online in conjunction with the teaching outcome feedback in the incremental teaching data to obtain the dynamic weight values.

[0048] Specifically, the steps for obtaining the initial weight values ​​of multidimensional evaluation indicators based on historical matching effect data include: extracting multidimensional evaluation indicators (knowledge transfer efficiency matching degree S1, learning style suitability S2, interaction participation fit S3, and teaching feedback responsiveness S4) from the target teacher-student combinations in historical matching schemes; and collecting student teaching outcome feedback data (such as grade improvement rate). Classroom participation (P). Calculate the Spearman rank correlation coefficients between each dimension of indicators and teaching outcomes. If a dimension indicator is significantly correlated with the outcome (p<0.05), its weight is retained; otherwise, it is discarded. The formula for assigning initial weight values ​​is: k = 1, 2, 3, 4. For example, Then initialize the weight values. , .

[0049] Specifically, the steps for obtaining dynamic weight values ​​by online calibration of the initial weight values ​​in the incremental teaching data, based on the feedback of teaching outcomes, include: dividing the incremental teaching data into data windows using a sliding window mechanism and aligning features with outcomes; using gradient descent optimization to online calibrate the initial weight values ​​to obtain dynamic weight values. For example, if the correlation between the interaction engagement fit S3 in the incremental teaching data and performance improvement increases, then the weight value of S3 increases from 0.19 to 0.25.

[0050] In S3, the steps to obtain the teacher-student matching evaluation results for the target teacher-student combination include: The teacher-student matching score of the target teacher-student combination is obtained by weighting and summing the specific quantitative values ​​and dynamic weight values ​​of the multidimensional evaluation indicators of the target teacher-student combination.

[0051] The teacher-student matching score is obtained by multiplying each specific quantitative value with its corresponding dynamic weight value and summing the results.

[0052] In one alternative approach, it also includes: Obtain teacher-student matching scores for multiple teacher-student pairs; A bipartite graph model for teacher-student matching is constructed, with students and teachers as nodes and teacher-student matching scores as edge weights. The minimum cost maximum flow algorithm is used to solve the bipartite graph model of teacher-student matching to obtain the globally optimal teacher-student pairing scheme.

[0053] The minimum-cost maximum flow algorithm is specifically a cost scaling algorithm. The globally optimal teacher-student pairing scheme must ensure that the number of students matched with each teacher does not exceed a threshold, and that every student is matched with a teacher; otherwise, a warning message is output.

[0054] The technical solution in this embodiment acquires multi-source heterogeneous data, cleans and standardizes it, constructs a multi-dimensional feature vector of the target teacher-student combination, and combines it with a matching degree prediction model and dynamic weight values ​​to achieve accurate quantitative evaluation of teacher-student matching degree. This embodiment makes full use of the massive data in the teaching process, breaks through the limitations of traditional manual matching, and solves problems such as single matching criteria and low efficiency. It can dynamically adapt to changes in the personalized characteristics of students and teachers, significantly improve the accuracy and scientific nature of matching, and provide an efficient and intelligent teacher-student matching solution for online education platforms, helping to realize personalized teaching and individualized instruction.

[0055] Figure 2 This diagram illustrates the structure of an embodiment of a teacher-student matching evaluation system 200 based on teaching big data analysis provided by the present invention. Figure 2 As shown, the system 200 includes: The data acquisition module 210 is used to acquire multi-source heterogeneous data generated by the target teacher-student combination during the teaching process. The multi-source heterogeneous data includes: students' classroom interaction behavior data, periodic assessment data, homework completion quality data, and teachers' teaching video feature data, course knowledge point coverage data, and historical student evaluation data. Processing module 220 is used to clean and standardize the multi-source heterogeneous data, and construct a multi-dimensional feature vector of the target teacher-student combination based on the processed multi-source heterogeneous data. The evaluation module 230 is used to input the multidimensional feature vector into the matching degree prediction model to obtain the specific quantitative values ​​of the multidimensional evaluation indicators of the target teacher-student combination, and generate dynamic weight values ​​of the multidimensional evaluation indicators based on historical matching effect data and incremental teaching data, so as to obtain the teacher-student matching degree evaluation result of the target teacher-student combination.

[0056] In one alternative approach, the multidimensional feature vector includes: a student multidimensional feature vector containing learning pattern preference features, knowledge mastery features, and cognitive level features; and a teacher multidimensional feature vector containing teaching style features, knowledge transfer efficiency features, and interactive response ability features.

[0057] In an alternative embodiment, the processing module 220 is specifically used for: The learning mode preference features are extracted from the classroom interaction behavior data, and the knowledge mastery features and cognitive level features are calculated based on the periodic assessment data and the homework completion quality data. The teaching style features are identified from the teaching video feature data, the knowledge transmission efficiency features are generated based on the course knowledge point coverage data, and the interactive response ability features are quantified through the historical student evaluation data.

[0058] In one alternative approach, the matching degree prediction model includes: a deep feature cross-network, a multi-head attention mechanism layer, a knowledge graph relationship reasoning module, a temporal behavior alignment model, a causal analysis model, and a multiple regression model; the multidimensional evaluation indicators include: knowledge transfer efficiency matching degree, learning style adaptation degree, interactive participation fit degree, and teaching feedback responsiveness degree.

[0059] In an alternative embodiment, the evaluation module 230 is specifically used for: The deep feature cross-network is used to perform nonlinear interactive operations on the student multidimensional feature vector and the teacher multidimensional feature vector to generate a high-order cross-feature matrix. The higher-order cross-feature matrix is ​​input into the multi-head attention mechanism layer. By calculating the correlation scores between the feature dimensions of the higher-order cross-feature matrix, feature dimension weights are generated. The higher-order cross-feature matrix is ​​then weighted and fused to obtain an optimized teacher-student interaction representation. The optimized teacher-student interaction representation is input into the knowledge graph relationship reasoning module to match the teacher's knowledge point coverage path with the student's knowledge mastery structure, thereby obtaining a specific quantitative value of the knowledge transmission efficiency matching degree. Learning pattern preference features and teaching style features are extracted from the optimized teacher-student interaction representation, and combined with the classroom interaction behavior data and the teaching video feature data, the dynamic matching value is calculated through the temporal behavior alignment model to obtain the specific quantitative value of the learning style fit. Based on the interactive response ability characteristics and cognitive level characteristics in the optimized teacher-student interaction representation, the causal analysis model is used to correlate historical interaction records with changes in assessment scores to obtain a specific quantitative value of the interaction participation fit. By integrating the interactive response capability characteristics in the optimized teacher-student interaction representation with historical student evaluation data, the impact of feedback timeliness on knowledge consolidation rate is predicted through the multivariate regression model, thereby obtaining a specific quantitative value of the teaching feedback responsiveness.

[0060] In an alternative embodiment, the evaluation module 230 is specifically used for: Based on the historical matching effect data, the initial weight values ​​of the multidimensional evaluation indicators are obtained, and the initial weight values ​​are calibrated online in conjunction with the teaching outcome feedback in the incremental teaching data to obtain the dynamic weight values.

[0061] In one optional approach, the teacher-student matching evaluation result is a teacher-student matching score; the evaluation module 230 is specifically used for: The teacher-student matching score of the target teacher-student combination is obtained by weighting and summing the specific quantitative values ​​and dynamic weight values ​​of the multidimensional evaluation indicators of the target teacher-student combination.

[0062] In an alternative embodiment, it further includes: a pairing module; the pairing module is used for: Obtain teacher-student matching scores for multiple teacher-student pairs; A bipartite graph model for teacher-student matching is constructed, with students and teachers as nodes and teacher-student matching scores as edge weights. The minimum cost maximum flow algorithm is used to solve the bipartite graph model of teacher-student matching to obtain the globally optimal teacher-student pairing scheme.

[0063] In one alternative approach, the cleaning and standardization process includes at least one of the following: data interpolation completion, outlier filtering, normalization, noise removal, and structuring.

[0064] The technical solution in this embodiment acquires multi-source heterogeneous data, cleans and standardizes it, constructs a multi-dimensional feature vector of the target teacher-student combination, and combines it with a matching degree prediction model and dynamic weight values ​​to achieve accurate quantitative evaluation of teacher-student matching degree. This embodiment makes full use of the massive data in the teaching process, breaks through the limitations of traditional manual matching, and solves problems such as single matching criteria and low efficiency. It can dynamically adapt to changes in the personalized characteristics of students and teachers, significantly improve the accuracy and scientific nature of matching, and provide an efficient and intelligent teacher-student matching solution for online education platforms, helping to realize personalized teaching and individualized instruction.

[0065] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

[0066] It should be noted that the terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and represent a limitation on a specific order or sequence. Where appropriate, the order of use for similar objects can be interchanged so that the embodiments of this application described herein can be implemented in an order other than that shown or described.

[0067] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1.A teacher-student matching degree evaluation method based on teaching big data analysis, characterized in that, The method comprises the following steps: acquiring multi-source heterogeneous data generated by a target teacher-student combination in a teaching process, the multi-source heterogeneous data comprising: student classroom interactive behavior data, stage test data, homework completion quality data, and teacher teaching video feature data, course knowledge point coverage data, and historical student evaluation data; cleaning and standardizing the multi-source heterogeneous data, and constructing a multi-dimensional feature vector of the target teacher-student combination according to the processed multi-source heterogeneous data; inputting the multi-dimensional feature vector into a matching degree prediction model to obtain specific quantitative values of multi-dimensional evaluation indexes of the target teacher-student combination, and generating dynamic weight values of the multi-dimensional evaluation indexes based on historical matching effect data and incremental teaching data to obtain a teacher-student matching degree evaluation result of the target teacher-student combination. 2.The teacher-student matching degree evaluation method based on teaching big data analysis according to claim 1, wherein, The multi-dimensional feature vector comprises: a student multi-dimensional feature vector comprising learning mode preference features, knowledge mastery features, and cognitive level hierarchy features, and a teacher multi-dimensional feature vector comprising teaching style features, knowledge transmission efficiency features, and interactive response capability features. 3.The teacher-student matching degree evaluation method based on teaching big data analysis according to claim 2, characterized in that, The step of constructing the multi-dimensional feature vector of the target teacher-student combination according to the processed multi-source heterogeneous data comprises: extracting the learning mode preference features from the classroom interactive behavior data, and calculating the knowledge mastery features and the cognitive level hierarchy features according to the stage test data and the homework completion quality data; identifying the teaching style features from the teaching video feature data, generating the knowledge transmission efficiency features based on the course knowledge point coverage data, and quantifying the interactive response capability features through the historical student evaluation data. 4.The teacher-student matching degree evaluation method based on teaching big data analysis according to claim 2, characterized in that, The matching degree prediction model comprises: a deep feature cross network, a multi-head attention mechanism layer, a knowledge graph relationship reasoning module, a time sequence behavior alignment model, a causal analysis model, and a multivariate regression model; and the multi-dimensional evaluation indexes comprise: knowledge transmission efficiency matching degree, learning style adaptation degree, interactive participation fit degree, and teaching feedback response degree. 5.The teacher-student matching degree evaluation method based on teaching big data analysis according to claim 4, characterized in that, The step of inputting the multi-dimensional feature vector into the matching degree prediction model to obtain specific quantitative values of multi-dimensional evaluation indexes of the target teacher-student combination comprises: performing nonlinear interaction operation on the student multi-dimensional feature vector and the teacher multi-dimensional feature vector by using the deep feature cross network to generate a high-order cross feature matrix; inputting the high-order cross feature matrix into the multi-head attention mechanism layer, generating feature dimension weights by calculating correlation scores between feature dimensions of the high-order cross feature matrix, and weighting and fusing the high-order cross feature matrix to obtain an optimized teacher-student interaction representation; inputting the optimized teacher-student interaction representation into the knowledge graph relationship reasoning module to match a teacher knowledge point coverage path and a student knowledge mastery structure to obtain a specific quantitative value of the knowledge transmission efficiency matching degree; extracting learning mode preference features and teaching style features from the optimized teacher-student interaction representation, and combining the classroom interactive behavior data and the teaching video feature data to calculate a dynamic matching value through the time sequence behavior alignment model to obtain a specific quantitative value of the learning style adaptation degree; Based on the interaction response ability feature and the cognitive level hierarchy feature in the optimized teacher-student interaction representation, the specific quantitative value of the interaction participation fit degree is obtained by using the causal analysis model to associate historical interaction records and test score changes; The specific quantitative value of the teaching feedback responsiveness is obtained by fusing the interaction response ability feature in the optimized teacher-student interaction representation and historical student evaluation data, and predicting the influence of feedback timeliness on knowledge consolidation rate through the multiple regression model. 6.The teacher-student matching degree evaluation method based on teaching big data analysis according to claim 4, characterized in that, The step of generating the dynamic weight value of the multi-dimensional evaluation index based on historical matching effect data and incremental teaching data includes: Based on the historical matching effect data, the initialization weight value of the multi-dimensional evaluation index is obtained, and the initialization weight value is calibrated online in combination with the teaching achievement feedback in the incremental teaching data, to obtain the dynamic weight value. 7.The teacher-student matching degree evaluation method based on teaching big data analysis of claim 1, wherein, The teacher-student matching degree evaluation result is a teacher-student matching degree score; The step of obtaining the teacher-student matching degree evaluation result of the target teacher-student combination includes: The specific quantitative value of the multi-dimensional evaluation index of the target teacher-student combination is weighted and summed with the dynamic weight value to obtain the teacher-student matching degree score of the target teacher-student combination. 8.The teacher-student matching degree evaluation method based on teaching big data analysis of claim 7, wherein, Also includes: Obtain the teacher-student matching degree scores of multiple pairs of teacher-student combinations; A teacher-student matching bipartite graph model is constructed with students and teachers as nodes and teacher-student matching degree scores as edge weights; The minimum cost maximum flow algorithm is used to solve the teacher-student matching bipartite graph model to obtain a globally optimal teacher-student pairing scheme. 9.The teacher-student matching degree evaluation method based on teaching big data analysis according to any one of claims 1 to 8, characterized in that, The cleaning and standardization processing mode includes at least one of data interpolation completion, outlier filtering, normalization processing, noise removal, and structured processing. 10.A teacher-student matching degree evaluation system based on teaching big data analysis, characterized in that, Includes: The acquisition module is configured to obtain multi-source heterogeneous data generated by the target teacher-student combination in the teaching process, wherein the multi-source heterogeneous data includes student classroom interaction behavior data, periodic test data, homework completion quality data, and teacher teaching video feature data, course knowledge point coverage data, and historical student evaluation data; The processing module is configured to clean and standardize the multi-source heterogeneous data, and construct a multi-dimensional feature vector of the target teacher-student combination based on the processed multi-source heterogeneous data: The evaluation module is configured to input the multi-dimensional feature vector into a matching degree prediction model to obtain a specific quantitative value of a multi-dimensional evaluation index of the target teacher-student combination, and generate a dynamic weight value of the multi-dimensional evaluation index based on historical matching effect data and incremental teaching data to obtain a teacher-student matching degree evaluation result of the target teacher-student combination.

Images