Education effect evaluation method and system based on artificial intelligence

Abstract

The invention discloses an educational effect evaluation method and system based on artificial intelligence. The method comprises the steps of multi-modal data preparation, dynamic cognition modeling, multi-dimensional effect evaluation, self-adaptive result optimization and educational effect evaluation. The invention relates to the technical field of education effect evaluation, and adopts a multi-modal data integration analysis and generative adversarial network and time sequence model fusion method to carry out real-time, long-period and migration education effect comprehensive evaluation, so that the performance of students in different learning stages can be evaluated more comprehensively and meticulously, and an accurate evaluation result is provided. A mode of combining cognitive network analysis and improving a dynamic Bayesian network is adopted, dynamic Bayesian inference and reinforcement learning optimization are utilized, and accurate modeling is carried out on a student cognitive process based on learning history and real-time feedback of students; and performing dynamic learning content and strategy recommendation on the students by adopting a mixed recommendation improvement method combined with group learning feature clustering.

Description

Technical Field

[0001] The present invention relates to the technical field of educational effect evaluation, and specifically refers to an educational effect evaluation method and system based on artificial intelligence. Background Art

[0002] The AI-based educational effectiveness evaluation method is an innovative way to intelligently analyze and evaluate students' learning behaviors, cognitive processes, and knowledge mastery by using technologies such as machine learning, natural language processing, and data mining. It collects multi-dimensional data (such as classroom performance, homework completion, interactive feedback, etc.) to build a personalized learning model, monitor learning progress in real time, and predict future performance. This method can not only accurately identify students' knowledge blind spots and learning preferences, but also provide teachers with scientific teaching optimization suggestions, thereby achieving teaching students in accordance with their aptitude and comprehensively improving teaching quality. Its core role is to transform traditional result-oriented evaluation into process-based and dynamic intelligent evaluation, and provide a data-driven scientific basis for educational decision-making.

[0003] However, in the existing educational effectiveness evaluation process, there are many traditional evaluation methods that only focus on one dimension (such as test scores or classroom interaction) to evaluate students' learning effects, and the existing evaluation methods often only focus on students' short-term effects, ignoring the technical problems of long-term learning effects and changes in deep knowledge mastery; in the existing dynamic cognitive modeling methods, there is a technical problem that the existing cognitive modeling methods are difficult to adapt to the dynamic changes of students' cognitive structure and strategies in the learning process, and the students' knowledge mastery in learning is constantly changing, and the traditional model cannot reflect these changes in real time; in the existing multidimensional effect evaluation methods, there is a technical problem that traditional evaluation dimensions often focus on one aspect (such as grades or participation). degree), ignoring other key factors in the learning process. Existing evaluation methods also tend to be short-term and immediate, which makes it difficult to effectively capture students' knowledge accumulation, thinking transfer ability and interdisciplinary comprehensive quality over a long period of time, resulting in a single evaluation dimension and the inability to fully reflect students' comprehensive learning ability. In the existing adaptive result optimization methods, there is a problem that the existing optimization methods usually lack learning content recommendations tailored to students' individual differences, and cannot provide accurate learning resources or strategies based on students' cognitive progress, learning status and behavioral characteristics. In addition, most traditional methods only consider the learning process of individual students, but ignore the potential of group learning characteristics, and lack personalized recommendations and group learning characteristics. Summary of the invention

[0004] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides an educational effect evaluation method and system based on artificial intelligence. In view of the fact that in the existing educational effect evaluation process, there are many traditional evaluation methods that only focus on a certain dimension (such as test scores or classroom interactions) to evaluate students' learning effects, and the existing evaluation methods often only focus on students' short-term effects, ignoring the technical problems of long-term learning effects and changes in deep knowledge mastery, this solution creatively adopts multimodal data integration analysis, combined with the fusion method of generative adversarial networks and time series models, to conduct a comprehensive evaluation of immediate, long-term and transfer educational effects, which can more comprehensively and meticulously evaluate students' performance at different learning stages, and provide Accurate evaluation results; In view of the technical problem that the existing dynamic cognitive modeling methods are difficult to adapt to the dynamic changes of students' cognitive structure and strategies in the learning process, and the students' knowledge mastery in learning is constantly changing, and the traditional model cannot reflect these changes in real time, this solution creatively adopts the method of combining cognitive network analysis with improved dynamic Bayesian network, using dynamic Bayesian inference and reinforcement learning optimization, and accurately models the students' cognitive process based on their learning history and real-time feedback. At the same time, the metacognitive adjustment module further improves the personalization and accuracy of the model through adaptive learning strategy optimization; In view of the technical problem that the existing multi-dimensional effect evaluation methods are difficult to adapt to the dynamic changes of students' cognitive structure and strategies in the learning process, and the students' knowledge mastery in learning is constantly changing, and the traditional model cannot reflect these changes in real time, this solution creatively adopts the method of combining cognitive network analysis with improved dynamic Bayesian network, using dynamic Bayesian inference and reinforcement learning optimization, and accurately models the students' cognitive process based on their learning history and real-time feedback. At the same time, the metacognitive adjustment module further improves the personalization and accuracy of the model through adaptive learning strategy optimization; As traditional assessments often focus on one aspect (such as grades or participation), ignoring other key factors in the learning process, existing assessment methods tend to be short-term and immediate, making it difficult to effectively capture students' knowledge accumulation, thinking transfer ability, and interdisciplinary comprehensive quality over a long period of time, resulting in a single assessment dimension and the inability to fully reflect students' comprehensive learning ability. This solution creatively adopts an innovative method based on the combination of generative adversarial networks and transformer timing networks. In the assessment process, it takes into account immediate feedback, long-term learning effects, and interdisciplinary transfer capabilities, ensuring that students' learning effects can be comprehensively evaluated from multiple dimensions, providing more accurate educational effect evaluation results; for the existing adaptive structure In the optimization methods, there is a problem that the existing optimization methods usually lack learning content recommendations tailored to individual differences among students, and cannot provide accurate learning resources or strategies based on students' cognitive process, learning status and behavioral characteristics. In addition, most traditional methods only consider the learning process of individual students, while ignoring the potential of group learning characteristics, and lack technical problems of personalized recommendation and group learning characteristics consideration. This scheme creatively adopts a hybrid recommendation improvement method combined with group learning feature clustering to dynamically recommend learning content and strategies to students, and through the analysis of students' cognitive process characteristics and group similarity data, provides a personalized and flexible learning process dynamic optimization plan, and achieves continuous improvement and optimization of educational effects.

[0005] The technical solution adopted by the present invention is as follows: The present invention provides an educational effect evaluation method based on artificial intelligence, the method comprising the following steps:

[0006] Step S1: multimodal data preparation;

[0007] Step S2: dynamic cognitive modeling;

[0008] Step S3: multi-dimensional effect evaluation;

[0009] Step S4: adaptive result optimization;

[0010] Step S5: Educational effectiveness evaluation.

[0011] Furthermore, in step S1, the multimodal data preparation is used to collect students' learning process data, specifically, by collecting the learning process data, collecting the original data of the students' learning process, and obtaining the multimodal education effect evaluation optimization data through knowledge construction analysis and data optimization;

[0012] The original data of the student learning process specifically includes classroom interaction record data, homework test data, classroom feedback data, student status evaluation data, classroom note data and extracurricular learning situation data;

[0013] The multimodal education effectiveness evaluation optimization data specifically includes optimizing classroom interaction data, optimizing student status data and optimizing learning process data.

[0014] Further, in step S2, the dynamic cognitive modeling is used to optimize the students' cognitive process through a dynamic modeling method, specifically, based on the multimodal education effect evaluation optimization data, a dynamic Bayesian network improved by combining cognitive network analysis and cognitive ability modeling is used to perform dynamic cognitive modeling and prediction to obtain cognitive process characteristic data, including the following steps:

[0015] Step S21: constructing original data input, specifically using the multimodal education effect evaluation optimization data as the original data input sample;

[0016] Step S22: cognitive network analysis, specifically, using a standard cognitive network analysis method to construct nodes and learning relationship paths, and based on the historical learning situation of students in the original data input sample, perform correlation analysis on the students' mastery of knowledge points to obtain knowledge point correlation reference data;

[0017] Step S23: constructing an improved dynamic Bayesian network, specifically constructing a standard dynamic Bayesian network and using the knowledge point relevance reference data as a state variable to perform improved dynamic Bayesian inference to obtain a basic modeling model of student knowledge point cognition;

[0018] Step S24: metacognitive adjustment, specifically, by constructing a student cognitive reward function, performing cognitive adjustment of knowledge points based on reinforcement learning, and by constructing a metacognitive ability model, simulating the student's adjustment of learning strategies, and by iterative training of reinforcement learning, obtaining a metacognitive correction student dynamic cognitive modeling model;

[0019] The student cognitive reward function is specifically composed of a weighted combination of a classroom interaction reward item, a homework test reward item, a classroom feedback reward item, a learning status reward item, and an in-class and extracurricular learning reward item;

[0020] Step S25: dynamic cognitive modeling, specifically, performing dynamic cognitive modeling through the construction of raw data input, the cognitive network analysis, the construction of an improved dynamic Bayesian network and the metacognitive adjustment to obtain cognitive process characteristic data;

[0021] The cognitive process characteristic data specifically includes knowledge point mastery data, learning strategy behavior data, learning emotional state change data, learning process prediction optimization data and meta-cognitive data;

[0022] The metacognitive data is used to indicate students' ability to monitor, evaluate and regulate their own learning process.

[0023] Further, in step S3, the multidimensional effect evaluation is used to evaluate the learning effect in multiple dimensions, specifically, based on the cognitive process characteristic data, a transformer timing network combined with generative adversarial transfer knowledge generation is used to perform an immediate, long-term and transfer three-dimensional educational effect evaluation to obtain multidimensional educational effect evaluation reference data, including the following steps:

[0024] Step S31: real-time education effect evaluation, specifically constructing a feedback-reward improved generative adversarial network to generate and evaluate the real-time education effect, and obtain real-time education effect evaluation reference data;

[0025] The feedback reward improvement is specifically designed by generating the student's learning progress data through a generator and calculating the expected probability of judgment;

[0026] Step S32: long-term education effect evaluation, specifically using a transformer timing network to perform long-term education effect evaluation modeling, and obtaining long-term education effect evaluation data through timing modeling;

[0027] Step S33: Transfer education effect evaluation, specifically, by combining the results of the instant education effect evaluation and transferring them to the interdisciplinary content, performing interdisciplinary transfer effect evaluation, and obtaining transfer education effect evaluation data;

[0028] Step S34: multidimensional effect evaluation model training, specifically, performing multidimensional effect evaluation model training through the instant education effect evaluation, the long-term education effect evaluation and the transfer education effect evaluation to obtain a multidimensional effect evaluation model;

[0029] Step S35: multidimensional effect evaluation, specifically, based on the cognitive process characteristic data, using the multidimensional effect evaluation model, to perform multidimensional effect evaluation to obtain multidimensional education effect evaluation reference data.

[0030] Further, in step S4, the adaptive result optimization is used to dynamically optimize the learning process according to the evaluation results, specifically, based on the cognitive process characteristic data and the multi-dimensional education effect evaluation reference data, a hybrid recommendation improvement method combined with group learning feature clustering is adopted to dynamically optimize the adaptive learning process, and obtain reference learning plan data for dynamic optimization of the learning process, including the following steps:

[0031] Step S41: group learning feature clustering, specifically, using a clustering algorithm to calculate student similarity based on the cognitive process feature data and the multi-dimensional education effect evaluation reference data to obtain similarity clustering learning feature data;

[0032] Step S42: recommending learning content, specifically recommending learning content based on the student's cognitive process data and the similarity clustering learning feature data to obtain learning content recommendation data;

[0033] Step S43: collaborative filtering recommendation, specifically, recommending learning strategies within a student group based on the multi-dimensional education effect evaluation reference data and the similarity clustering learning feature data, and obtaining intra-group recommendation data;

[0034] Step S44: integrating learning content recommendations, specifically weighting and integrating the learning content recommendation data and the intra-group recommendation data to obtain balanced learning content recommendation reference data;

[0035] Step S45: Adaptive result optimization, specifically, testing and adjusting the weights of the learning content recommendation integration to obtain reference learning plan data for dynamic optimization of the learning process.

[0036] Furthermore, in step S5, the educational effect evaluation is used to integrate the processes of cognitive modeling, effect evaluation and result optimization and conduct a comprehensive evaluation of educational effects. Specifically, the reference learning plan data for dynamic optimization of the learning process and the multidimensional educational effect evaluation reference data are combined to conduct educational effect evaluation and personalized learning plan optimization for students to obtain comprehensive reference data for educational effect evaluation.

[0037] The present invention provides an artificial intelligence-based education effect evaluation system, comprising a multimodal data preparation module, a dynamic cognitive modeling module, a multidimensional effect evaluation module, an adaptive result optimization module and an education effect evaluation module;

[0038] The multimodal data preparation module is used for preparing multimodal data, obtaining multimodal education effect evaluation optimization data through multimodal data preparation, and sending the multimodal education effect evaluation optimization data to the dynamic cognitive modeling module;

[0039] The dynamic cognitive modeling module is used for dynamic cognitive modeling, obtains cognitive process characteristic data through dynamic cognitive modeling, and sends the cognitive process characteristic data to the multi-dimensional effect evaluation module and the adaptive result optimization module;

[0040] The multidimensional effect evaluation module is used for multidimensional effect evaluation, obtains multidimensional education effect evaluation reference data through multidimensional effect evaluation, and sends the multidimensional education effect evaluation reference data to the adaptive result optimization module and the education effect evaluation module;

[0041] The adaptive result optimization module is used for adaptive result optimization, and obtains reference learning plan data for dynamic optimization of the learning process through adaptive result optimization, and sends the reference learning plan data for dynamic optimization of the learning process to the education effect evaluation module;

[0042] The educational effect evaluation module is used for educational effect evaluation, and through the educational effect evaluation, comprehensive reference data for educational effect evaluation is obtained.

[0043] The beneficial effects achieved by the present invention using the above scheme are as follows:

[0044] (1) In the existing educational effect evaluation process, there are many traditional evaluation methods that only focus on one dimension (such as test scores or classroom interactions) to evaluate students' learning effects. In addition, existing evaluation methods often only focus on students' short-term effects and ignore the technical problems of long-term learning effects and changes in deep knowledge mastery. This solution creatively adopts multimodal data integration analysis, combined with the fusion method of generative adversarial networks and time series models, to conduct a comprehensive evaluation of immediate, long-term and transfer education effects. It can more comprehensively and meticulously evaluate students' performance at different learning stages and provide accurate evaluation results.

[0045] (2) In view of the technical problem that the existing dynamic cognitive modeling methods are difficult to adapt to the dynamic changes of students' cognitive structure and strategies in the learning process, and the students' knowledge mastery in learning is constantly changing, and the traditional models cannot reflect these changes in real time, this solution creatively adopts the method of combining cognitive network analysis with improved dynamic Bayesian network, using dynamic Bayesian inference and reinforcement learning optimization, and accurately models the students' cognitive process based on their learning history and real-time feedback. At the same time, the metacognitive adjustment module further improves the personalization and accuracy of the model through adaptive learning strategy optimization;

[0046] (3) In view of the fact that in the existing multi-dimensional effect evaluation methods, traditional evaluation dimensions often focus on one aspect (such as grades or participation), ignoring other key factors in the learning process, and existing evaluation methods tend to be short-term and immediate evaluation, which makes it difficult to effectively capture students' knowledge accumulation, thinking transfer ability and interdisciplinary comprehensive quality over a long period of time, resulting in a single evaluation dimension and the inability to fully reflect students' comprehensive learning ability. This solution creatively adopts an innovative method based on the combination of generative adversarial networks and transformer timing networks. In the evaluation process, it takes into account immediate feedback, long-term learning effects and interdisciplinary transfer ability, ensuring that students' learning effects can be comprehensively evaluated from multiple dimensions and providing more accurate educational effect evaluation results;

[0047] (4) In view of the fact that existing adaptive result optimization methods usually lack learning content recommendations tailored to individual students’ differences, and are unable to provide accurate learning resources or strategies based on students’ cognitive process, learning status, and behavioral characteristics, and that most traditional methods only consider the learning process of individual students, while ignoring the potential of group learning characteristics, and lack technical issues such as the lack of personalized recommendations and group learning characteristics, this solution creatively adopts a hybrid recommendation improvement method combined with group learning feature clustering to dynamically recommend learning content and strategies to students, and by analyzing students’ cognitive process characteristics and group similarity data, provides a personalized and flexible dynamic optimization plan for the learning process, thereby achieving continuous improvement and optimization of educational results. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A flowchart of an artificial intelligence-based educational effect evaluation method provided by the present invention;

[0049] Figure 2 A schematic diagram of an education effect evaluation system based on artificial intelligence provided by the present invention;

[0050] Figure 3 A schematic diagram of the process of dynamic cognitive modeling in step S2;

[0051] Figure 4 This is a flow chart of multi-dimensional effect evaluation in step S3;

[0052] Figure 5 Schematic diagram of the process of optimizing the adaptive results in step S4.

[0053] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention. DETAILED DESCRIPTION

[0054] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments; based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0055] In the description of the present invention, it should be understood that terms such as “upper”, “lower”, “front”, “back”, “left”, “right”, “top”, “bottom”, “inside” and “outside” indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore should not be understood as limiting the present invention.

[0056] Example 1, see Figure 1 The present invention provides an artificial intelligence-based educational effect evaluation method, which comprises the following steps:

[0057] Step S1: multimodal data preparation;

[0058] Step S2: dynamic cognitive modeling;

[0059] Step S3: multi-dimensional effect evaluation;

[0060] Step S4: adaptive result optimization;

[0061] Step S5: Educational effectiveness evaluation.

[0062] By performing the above operations, in view of the fact that in the existing educational effect evaluation process, there are many traditional evaluation methods that only focus on one dimension (such as test scores or classroom interactions) to evaluate students' learning effects, and the existing evaluation methods often only focus on students' short-term effects, ignoring the technical problems of long-term learning outcomes and changes in deep knowledge mastery, this solution creatively adopts multimodal data integration analysis, combined with the fusion method of generative adversarial networks and time series models, to conduct a comprehensive evaluation of immediate, long-term and transfer education effects, which can more comprehensively and meticulously evaluate students' performance at different learning stages and provide accurate evaluation results.

[0063] Example 2, see Figure 1 and Figure 2 In step S1, the multimodal data preparation is used to collect students' learning process data, specifically, by collecting the learning process data, collecting the original data of the students' learning process, and obtaining the multimodal education effect evaluation optimization data through knowledge construction analysis and data optimization;

[0064] The original data of the student learning process specifically includes classroom interaction record data, homework test data, classroom feedback data, student status evaluation data, classroom note data and extracurricular learning situation data;

[0065] The classroom interaction record data is collected through manual recording, including data on the number of questions asked by students, data on students’ answers, and classroom performance record data; the classroom performance record data includes the number of active speeches, the accuracy of answers, and manual evaluation data on classroom concentration;

[0066] The homework test data is collected through manual recording, including homework completion status and test score status data;

[0067] The classroom feedback data is collected by filling out a questionnaire, including student self-feedback data and student emotional feedback data;

[0068] The student status assessment data is collected through manual recording, including manual assessment status data and manual assessment student emotion data;

[0069] The class notes data is collected through manual recording and questionnaire filling, including data on whether students take notes and whether there is class discussion;

[0070] The extracurricular learning data is collected by filling out questionnaires, including extracurricular learning record data and student self-study record data

[0071] The knowledge construction analysis and data optimization specifically include labeling analysis and data optimization of the original data of the student learning process to obtain knowledge construction analysis data;

[0072] The multimodal education effectiveness evaluation optimization data specifically includes optimizing classroom interaction data, optimizing student status data and optimizing learning process data.

[0073] Example 3, see Figure 1 , Figure 2 and Figure 3 This embodiment is based on the above embodiment. In step S2, the dynamic cognitive modeling is used to optimize the students' cognitive process through a dynamic modeling method. Specifically, based on the multimodal education effect evaluation optimization data, a dynamic Bayesian network improved by combining cognitive network analysis and cognitive ability modeling is used to perform dynamic cognitive modeling and prediction to obtain cognitive process characteristic data, including the following steps:

[0074] Step S21: constructing original data input, specifically using the multimodal education effect evaluation optimization data as the original data input sample;

[0075] Step S22: cognitive network analysis, specifically using a standard cognitive network analysis method to construct knowledge nodes and learning relationship paths, and based on the historical learning situation of students in the original data input sample, perform correlation analysis on the degree of students' mastery of knowledge points to obtain knowledge point correlation reference data;

[0076] Step S23: constructing an improved dynamic Bayesian network, specifically constructing a standard dynamic Bayesian network and using the knowledge point relevance reference data as a state variable to perform improved dynamic Bayesian inference to obtain a basic modeling model of student knowledge point cognition;

[0077] The calculation formula of the dynamic Bayesian inference for constructing a standard dynamic Bayesian network and improving the knowledge point association degree reference data as a state variable is:

[0078] ;

[0079] In the formula, P(c t |c t-1 ,X t ) is the dynamic Bayesian inference probability of cognitive state, which is used to indicate the degree of students’ cognition of the current knowledge point. t is the cognitive state variable, X t is the student cognitive behavior data, used to represent the knowledge point relevance reference data, a is the normalization constant, P(c t |c t-1 ) is the overall probability of cognitive state influence, which is used to indicate the degree of influence of the student's cognitive state of the knowledge point at time t-1 on the current time t. t |c t) is the overall probability of student cognitive behavior influence, which is used to indicate the degree to which students’ cognitive behavior is influenced by cognitive state variables;

[0080] Step S24: metacognitive adjustment, specifically, by constructing a student cognitive reward function, performing cognitive adjustment of knowledge points based on reinforcement learning, and by constructing a metacognitive ability model, simulating the student's adjustment of learning strategies, and by iterative training of reinforcement learning, obtaining a metacognitive correction student dynamic cognitive modeling model;

[0081] The method comprises: optimizing the basic modeling model of student knowledge point cognition by using an improved metacognitive correction factor through iterative training of reinforcement learning, obtaining a metacognitive adjustment model of student knowledge points, constructing a student cognitive reward function, and performing reinforcement learning;

[0082] The calculation formula of the student knowledge point metacognitive adjustment model is:

[0083] ;

[0084] In the formula, The overall result is the dynamic Bayesian inference probability of cognitive state after meta-cognitive adjustment of students’ knowledge points. It is the metacognitive adjustment factor, which is used to represent students’ self-adjustment and feedback on their learning situation;

[0085] The student cognitive reward function is specifically composed of the weighted items of classroom interaction reward, homework test reward, classroom feedback reward, learning status reward, and in-class and extracurricular learning reward. The calculation formula is:

[0086] ;

[0087] Where R(t) is the student cognitive reward function, w1 is the classroom interaction reward weight, and R inter (t) is the sub-item function of classroom interaction reward, w2 is the homework test reward weight, R test (t) is the sub-item function of homework and test rewards, w3 is the weight of classroom feedback rewards, R febk (t) is the sub-item function of classroom feedback reward, w4 is the learning state reward weight, R notes (t) is the learning status reward function, w5 is the weight of the in-class and extracurricular learning reward, R extra (t) is the sub-item function of in-class and extracurricular learning rewards;

[0088] Table 1 is a table showing the composition of the sub-item reward function of the student cognitive reward function. As shown in Table 1, the sub-item function R of the classroom interaction reward is inter(t), a1 is the weight of the number of questions asked, Q(t) is the number of questions asked by students, a2 is the weight of the correct rate of students’ answers, P(t) is the correct rate of students’ answers, a3 is the weight of students’ class concentration, and C(t) is the class concentration term;

[0089] The assignment test reward item function R test (t), b1 is the weight of the homework completion, A(t) is the homework completion item, b2 is the weight of the test score, S(t) is the test score item

[0090] The classroom feedback reward item function R febk (t), c1 is the weight of the self-feedback score, F s (t) is the self-feedback score item, c2 is the emotional feedback score weight, F e (t) is the emotional feedback score item;

[0091] The learning state reward sub-item function R notes In (t), d1 is the weight of whether the student takes notes or not, E N (t) is the student note-taking item, d2 is the weight of class discussion, E D (t) whether it is a class discussion or not;

[0092] The in-class and extracurricular learning reward item function R extra In (t), e1 is the weight of extracurricular learning time, L ss (t) is the extracurricular study time item, e2 is the weight of the course study time, L ex (t) is the duration of course study and self-study;

[0093] Table 1 Composition of the sub-item reward function of the student cognitive reward function

[0094]

[0095] Step S25: dynamic cognitive modeling, specifically, performing dynamic cognitive modeling through the construction of raw data input, the cognitive network analysis, the construction of an improved dynamic Bayesian network and the metacognitive adjustment to obtain cognitive process characteristic data;

[0096] The cognitive process characteristic data specifically includes knowledge point mastery data, learning strategy behavior data, learning emotional state change data, learning process prediction optimization data and meta-cognitive data;

[0097] The metacognitive data is used to indicate students' ability to monitor, evaluate and regulate their own learning process.

[0098] By performing the above operations, in view of the technical problem that the existing dynamic cognitive modeling methods are difficult to adapt to the dynamic changes of students' cognitive structure and strategies in the learning process, and the students' knowledge mastery in learning is constantly changing, and the traditional models cannot reflect these changes in real time, this solution creatively adopts the method of combining cognitive network analysis with improved dynamic Bayesian network, using dynamic Bayesian inference and reinforcement learning optimization, and accurately models the students' cognitive process based on their learning history and real-time feedback. At the same time, the metacognitive adjustment module further improves the personalization and accuracy of the model through adaptive learning strategy optimization.

[0099] Example 4, see Figure 1 , Figure 2 and Figure 4 This embodiment is based on the above embodiment. In step S3, the multi-dimensional effect evaluation is used to evaluate the learning effect in multiple dimensions. Specifically, based on the cognitive process feature data, a transformer timing network combined with generative adversarial transfer knowledge generation is used to perform three-dimensional education effect evaluation of immediate, long-term and transfer, and obtain multi-dimensional education effect evaluation reference data, including the following steps:

[0100] Step S31: real-time education effect evaluation, specifically constructing a feedback-reward improved generative adversarial network to generate and evaluate the real-time education effect, and obtain real-time education effect evaluation reference data;

[0101] The feedback reward improvement is specifically designed by generating the student's learning progress data through a generator and calculating the expected probability of judgment;

[0102] The calculation formula for the feedback reward improvement design is:

[0103] ;

[0104] In the formula, R instant (t) is a feedback reward improvement function, which is used to calculate the expected probability of generation judgment of learning progress data. By maximizing the feedback reward improvement function, it is used as the optimization goal of the real-time education effect evaluation. real [·] is the expected probability of judging that it is true, E fake [·] is the expected probability of being judged as false, x t is the actual learning progress data, is the generated predicted learning progress data, D real (·) is the probability of the discriminator judging the real data, D fake (·) is the probability of the discriminator’s decision on the generated data;

[0105] Step S32: long-term education effect evaluation, specifically using a transformer timing network to perform long-term education effect evaluation modeling, and obtaining long-term education effect evaluation data through timing modeling;

[0106] The calculation formula for the long-term education effect evaluation is:

[0107] ;

[0108] In the formula, R long (t) is the long-term learning reward function, which is used to calculate the learning benefits over time t and serves as the optimization target for long-term education effect evaluation. T is the total time, t is the time index, is a nonlinear conversion function, n is the total number of long-term education effect evaluation tasks, i is the index of the long-term education effect evaluation task, is the weight of the i-th educational effect evaluation task, S test (·) is the long-term education effect evaluation function, which is used to represent the learning effect of students in long-term learning. The whole is the time decay factor, is the time decay coefficient;

[0109] Step S33: Transfer education effect evaluation, specifically, by combining the results of the instant education effect evaluation and transferring them to the interdisciplinary content, performing interdisciplinary transfer effect evaluation, and obtaining transfer education effect evaluation data;

[0110] The interdisciplinary migration effect evaluation is migrated and constructed in combination with the feedback improvement of the instant education effect evaluation. The calculation formula is:

[0111] ;

[0112] In the formula, R transfer (t) is the interdisciplinary transfer effect evaluation reward function, which is used to calculate the adversarial loss between the source domain and the target domain. By minimizing the feedback reward improvement function, it is used as the optimization target of the interdisciplinary transfer effect evaluation reward function. source [·] is the adversarial loss in the source domain, E target [·] is the target domain, X s It is the source domain data, which is used to represent the original data before cross-disciplinary migration. is the target domain data, which is used to represent the data after the source migration. source (·) is the source domain output, D target (·) is the target domain output;

[0113] Step S34: multidimensional effect evaluation model training, specifically, performing multidimensional effect evaluation model training through the instant education effect evaluation, the long-term education effect evaluation and the transfer education effect evaluation to obtain a multidimensional effect evaluation model;

[0114] Step S35: multidimensional effect evaluation, specifically, based on the cognitive process characteristic data, using the multidimensional effect evaluation model, to perform multidimensional effect evaluation to obtain multidimensional education effect evaluation reference data.

[0115] By performing the above operations, in view of the fact that in the existing multi-dimensional effect evaluation methods, traditional evaluation dimensions often focus on one aspect (such as grades or participation), ignoring other key factors in the learning process, and the existing evaluation methods are also biased towards short-term instant evaluation, which is difficult to effectively capture students' knowledge accumulation, thinking transfer ability and interdisciplinary comprehensive quality over a long period of time, resulting in a single evaluation dimension and the inability to fully reflect the technical problem of students' comprehensive learning ability. This solution creatively adopts an innovative method based on the combination of generative adversarial networks and transformer timing networks. In the evaluation process, instant feedback, long-term learning effects and interdisciplinary transfer capabilities are taken into account to ensure that students' learning effects can be comprehensively evaluated from multiple dimensions, providing more accurate education effect evaluation results.

[0116] Example 5, see Figure 1 , Figure 2 and Figure 5 This embodiment is based on the above embodiment. In step S4, the adaptive result optimization is used to dynamically optimize the learning process according to the evaluation results. Specifically, based on the cognitive process feature data and the multi-dimensional education effect evaluation reference data, a hybrid recommendation improvement method combined with group learning feature clustering is adopted to dynamically optimize the adaptive learning process to obtain reference learning plan data for dynamic optimization of the learning process, including the following steps:

[0117] Step S41: group learning feature clustering, specifically, using a clustering algorithm to calculate student similarity based on the cognitive process feature data and the multi-dimensional education effect evaluation reference data to obtain similarity clustering learning feature data;

[0118] The student similarity calculation specifically uses K-means clustering algorithm to perform similarity calculation;

[0119] Step S42: Recommending learning content, specifically, recommending learning content based on the student's cognitive process data and the similarity clustering learning feature data to obtain learning content recommendation data, and the calculation formula is:

[0120] ;

[0121] In the formula, It is the learning content recommendation data, which is used to indicate the recommendation to students. The rating value of the recommended learning content c, is the student index, c is the learning content index, K is the total number of similarity clustering learning features, k is the similarity clustering learning feature index, and w k is the feature weight of similarity clustering learning, sim(·) is the similarity function, Are a student The value of similarity clustering learning features, is the value of the learning content on the similarity clustering learning feature;

[0122] Step S43: collaborative filtering recommendation, specifically, recommending learning strategies within the student group based on the multi-dimensional education effect evaluation reference data and the similarity clustering learning feature data, and obtaining the recommended data within the group, and the calculation formula is:

[0123] ;

[0124] In the formula, It is the recommendation data within the group, which is used to indicate the recommendation to students. The collaborative filtering score of the recommended learning content c, Are a student The total number of neighboring students, is the neighbor student index, A neighboring student For the historical reference score of learning content c, sim(·) is the similarity function, Are a student Neighboring students similarities between;

[0125] Step S44: integrating learning content recommendations, specifically weighting and integrating the learning content recommendation data and the intra-group recommendation data to obtain balanced learning content recommendation reference data, and the calculation formula is:

[0126] ;

[0127] In the formula, R is the reference data for the recommended balanced learning content. is the learning content recommendation weight, is the learning content recommendation data, It is the recommendation data within the group;

[0128] Step S45: Adaptive result optimization, specifically, testing and adjusting the weights of the learning content recommendation integration to obtain reference learning plan data for dynamic optimization of the learning process.

[0129] By performing the above operations, in view of the fact that in the existing adaptive result optimization methods, the existing optimization methods usually lack learning content recommendations based on individual differences of students, and cannot provide accurate learning resources or strategies according to students' cognitive process, learning status and behavioral characteristics, and most traditional methods only consider the learning process of individual students, but ignore the potential of group learning characteristics, and lack the technical problems of personalized recommendation and group learning characteristics consideration, this scheme creatively adopts a hybrid recommendation improvement method combined with group learning feature clustering to dynamically recommend learning content and strategies to students, and through the analysis of students' cognitive process characteristics and group similarity data, provides a personalized and flexible learning process dynamic optimization scheme, and achieves continuous improvement and optimization of educational effects.

[0130] Example 6, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. In step S5, the educational effect evaluation is used to integrate the processes of cognitive modeling, effect evaluation and result optimization and conduct a comprehensive evaluation of the educational effect. Specifically, the learning process is dynamically optimized with reference to the learning plan data and the multi-dimensional educational effect evaluation reference data to conduct educational effect evaluation and personalized learning plan optimization for students, and obtain comprehensive reference data for educational effect evaluation.

[0131] Embodiment 7, see Figure 1 and Figure 2 , this embodiment is based on the above embodiment, and the present invention provides an education effect evaluation system based on artificial intelligence, including a multimodal data preparation module, a dynamic cognitive modeling module, a multidimensional effect evaluation module, an adaptive result optimization module and an education effect evaluation module;

[0132] The multimodal data preparation module is used for preparing multimodal data, obtaining multimodal education effect evaluation optimization data through multimodal data preparation, and sending the multimodal education effect evaluation optimization data to the dynamic cognitive modeling module;

[0133] The dynamic cognitive modeling module is used for dynamic cognitive modeling, obtains cognitive process characteristic data through dynamic cognitive modeling, and sends the cognitive process characteristic data to the multi-dimensional effect evaluation module and the adaptive result optimization module;

[0134] The multidimensional effect evaluation module is used for multidimensional effect evaluation, obtains multidimensional education effect evaluation reference data through multidimensional effect evaluation, and sends the multidimensional education effect evaluation reference data to the adaptive result optimization module and the education effect evaluation module;

[0135] The adaptive result optimization module is used for adaptive result optimization, and obtains reference learning plan data for dynamic optimization of the learning process through adaptive result optimization, and sends the reference learning plan data for dynamic optimization of the learning process to the education effect evaluation module;

[0136] The educational effect evaluation module is used for educational effect evaluation, and through the educational effect evaluation, comprehensive reference data for educational effect evaluation is obtained.

[0137] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0138] While the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that many changes, modifications, substitutions and variations can be made to the embodiments without departing from the principles and spirit of the invention.

[0139] The present invention and its embodiments are described above, and such description is not restrictive. The drawings show only one embodiment of the present invention, and the actual structure is not limited thereto. In short, if ordinary technicians in the field are inspired by it, without departing from the purpose of the invention, they can design a structure and embodiment similar to the technical solution without creativity, which should belong to the protection scope of the present invention.

Claims

1. An educational effect evaluation method based on artificial intelligence, characterized by: The method comprises the following steps: Step S1: multimodal data preparation to obtain multimodal education effect evaluation optimization data; Step S2: Dynamic cognitive modeling, using a dynamic Bayesian network improved by combining cognitive network analysis and cognitive ability modeling to perform dynamic cognitive modeling and prediction, and obtain cognitive process characteristic data; The improved dynamic Bayesian network for cognitive modeling is specifically to construct a standard dynamic Bayesian network and use the knowledge point relevance reference data as a state variable to perform improved dynamic Bayesian inference to obtain a basic modeling model for student knowledge point cognition; The dynamic cognitive modeling and prediction is carried out by constructing a student cognitive reward function to adjust the knowledge point cognition based on reinforcement learning; the student cognitive reward function is specifically composed of a weighted combination of classroom interaction reward items, homework test reward items, classroom feedback reward items, learning status reward items, and in-class and extracurricular learning reward items, and the calculation formula is: ； Where R(t) is the student cognitive reward function, w1 is the classroom interaction reward weight, and R inter (t) is the sub-item function of classroom interaction reward, w2 is the homework test reward weight, R test (t) is the sub-item function of homework and test rewards, w3 is the weight of classroom feedback rewards, R febk (t) is the sub-item function of classroom feedback reward, w4 is the learning state reward weight, R notes (t) is the learning status reward function, w5 is the weight of the in-class and extracurricular learning reward, R extra (t) is the sub-item function of in-class and extracurricular learning rewards; Step S3: Multi-dimensional effect evaluation, using a transformer-based time series network generated by generative adversarial transfer knowledge to conduct three-dimensional educational effect evaluations of immediate, long-term, and transfer, and obtain multi-dimensional educational effect evaluation reference data; Step S4: Adaptive result optimization, using a hybrid recommendation improvement method combined with group learning feature clustering to dynamically optimize the adaptive learning process, and obtain reference learning plan data for dynamic optimization of the learning process; Step S5: Evaluate the educational effect and obtain comprehensive reference data for the educational effect evaluation.

2. The method for evaluating educational effects based on artificial intelligence according to claim 1, characterized in that: In step S1, the multimodal data preparation is used to collect students' learning process data, specifically, by collecting the learning process data, collecting the original data of the students' learning process, and obtaining the multimodal education effect evaluation optimization data through knowledge construction analysis and data optimization; The original data of the student learning process specifically includes classroom interaction record data, homework test data, classroom feedback data, student status evaluation data, classroom note data and extracurricular learning situation data; The multimodal education effectiveness evaluation optimization data specifically includes optimizing classroom interaction data, optimizing student status data and optimizing learning process data.

3. The method for evaluating educational effects based on artificial intelligence according to claim 2, characterized in that: In step S2, the dynamic cognitive modeling is used to optimize the students' cognitive process through a dynamic modeling method, specifically, based on the multimodal education effect evaluation optimization data, a dynamic Bayesian network improved by combining cognitive network analysis and cognitive ability modeling is used to perform dynamic cognitive modeling and prediction to obtain cognitive process characteristic data, including the following steps: Step S21: constructing original data input, specifically using the multimodal education effect evaluation optimization data as the original data input sample; Step S22: cognitive network analysis, specifically, using a standard cognitive network analysis method to construct nodes and learning relationship paths, and based on the historical learning situation of students in the original data input sample, perform correlation analysis on the students' mastery of knowledge points to obtain knowledge point correlation reference data; Step S23: constructing an improved dynamic Bayesian network, specifically constructing a standard dynamic Bayesian network and using the knowledge point relevance reference data as a state variable to perform improved dynamic Bayesian inference to obtain a basic modeling model of student knowledge point cognition; Step S24: metacognitive adjustment, specifically, by constructing a student cognitive reward function, performing cognitive adjustment of knowledge points based on reinforcement learning, and by constructing a metacognitive ability model, simulating the student's adjustment of learning strategies, and by iterative training of reinforcement learning, obtaining a metacognitive correction student dynamic cognitive modeling model; The method comprises: optimizing the basic modeling model of student knowledge point cognition by using an improved metacognitive correction factor through iterative training of reinforcement learning, obtaining a metacognitive adjustment model of student knowledge points, constructing a student cognitive reward function, and performing reinforcement learning; The student cognitive reward function is specifically composed of a weighted combination of a classroom interaction reward item, a homework test reward item, a classroom feedback reward item, a learning status reward item, and an in-class and extracurricular learning reward item; Step S25: Dynamic cognitive modeling, specifically, performing dynamic cognitive modeling through the construction of original data input, the cognitive network analysis, the construction of an improved dynamic Bayesian network and the metacognitive adjustment to obtain cognitive process characteristic data.

4. The method for evaluating educational effects based on artificial intelligence according to claim 3, characterized in that: In step S2, the cognitive process characteristic data specifically includes knowledge point mastery data, learning strategy behavior data, learning emotional state change data, learning process prediction optimization data and metacognitive data; The metacognitive data is used to indicate students' ability to monitor, evaluate and regulate their own learning process.

5. The method for evaluating educational effects based on artificial intelligence according to claim 4, characterized in that: In step S3, the multidimensional effect evaluation is used to evaluate the learning effect in multiple dimensions. Specifically, based on the cognitive process characteristic data, a transformer timing network combined with generative adversarial transfer knowledge generation is used to perform an immediate, long-term and transfer three-dimensional educational effect evaluation to obtain multidimensional educational effect evaluation reference data, including the following steps: Step S31: real-time education effect evaluation, specifically constructing a feedback-reward improved generative adversarial network to generate and evaluate the real-time education effect, and obtain real-time education effect evaluation reference data; The feedback reward improvement is specifically designed by generating the student's learning progress data through a generator and calculating the expected probability of judgment; Step S32: long-term education effect evaluation, specifically using a transformer timing network to perform long-term education effect evaluation modeling, and obtaining long-term education effect evaluation data through timing modeling; Step S33: Transfer education effect evaluation, specifically, by combining the results of the instant education effect evaluation and transferring them to the interdisciplinary content, performing interdisciplinary transfer effect evaluation, and obtaining transfer education effect evaluation data; Step S34: multidimensional effect evaluation model training, specifically, performing multidimensional effect evaluation model training through the instant education effect evaluation, the long-term education effect evaluation and the transfer education effect evaluation to obtain a multidimensional effect evaluation model; Step S35: multidimensional effect evaluation, specifically, based on the cognitive process characteristic data, using the multidimensional effect evaluation model, to perform multidimensional effect evaluation to obtain multidimensional education effect evaluation reference data.

6. The method for evaluating educational effects based on artificial intelligence according to claim 5, characterized in that: In step S4, the adaptive result optimization is used to dynamically optimize the learning process according to the evaluation results, specifically, based on the cognitive process feature data and the multi-dimensional education effect evaluation reference data, a hybrid recommendation improvement method combined with group learning feature clustering is used to dynamically optimize the adaptive learning process to obtain reference learning plan data for dynamic optimization of the learning process, including the following steps: Step S41: group learning feature clustering, specifically, using a clustering algorithm to calculate student similarity based on the cognitive process feature data and the multi-dimensional education effect evaluation reference data to obtain similarity clustering learning feature data; Step S42: recommending learning content, specifically recommending learning content based on the student's cognitive process data and the similarity clustering learning feature data to obtain learning content recommendation data; Step S43: collaborative filtering recommendation, specifically, recommending learning strategies within a student group based on the multi-dimensional education effect evaluation reference data and the similarity clustering learning feature data, and obtaining intra-group recommendation data; Step S44: integrating learning content recommendations, specifically weighting and integrating the learning content recommendation data and the intra-group recommendation data to obtain balanced learning content recommendation reference data, and the calculation formula is: ； In the formula, R is the reference data for the recommended balanced learning content. is the learning content recommendation weight, is the learning content recommendation data, It is the recommendation data within the group; Step S45: Adaptive result optimization, specifically, testing and adjusting the weights of the learning content recommendation integration to obtain reference learning plan data for dynamic optimization of the learning process.

7. The method for evaluating educational effects based on artificial intelligence according to claim 6, characterized in that: In step S5, the educational effect evaluation is used to integrate the processes of cognitive modeling, effect evaluation and result optimization and conduct a comprehensive evaluation of the educational effect. Specifically, the reference learning plan data for dynamic optimization of the learning process and the multidimensional educational effect evaluation reference data are combined to conduct educational effect evaluation and personalized learning plan optimization for students to obtain comprehensive reference data for educational effect evaluation.

8. An education effect evaluation system based on artificial intelligence, used to implement an education effect evaluation method based on artificial intelligence as described in any one of claims 1 to 7, characterized in that: It includes multimodal data preparation module, dynamic cognitive modeling module, multidimensional effect evaluation module, adaptive result optimization module and educational effect evaluation module; The multimodal data preparation module is used for preparing multimodal data, obtaining multimodal education effect evaluation optimization data through multimodal data preparation, and sending the multimodal education effect evaluation optimization data to the dynamic cognitive modeling module; The dynamic cognitive modeling module is used for dynamic cognitive modeling, obtains cognitive process characteristic data through dynamic cognitive modeling, and sends the cognitive process characteristic data to the multi-dimensional effect evaluation module and the adaptive result optimization module; The multidimensional effect evaluation module is used for multidimensional effect evaluation, obtains multidimensional education effect evaluation reference data through multidimensional effect evaluation, and sends the multidimensional education effect evaluation reference data to the adaptive result optimization module and the education effect evaluation module; The adaptive result optimization module is used for adaptive result optimization, and obtains reference learning plan data for dynamic optimization of the learning process through adaptive result optimization, and sends the reference learning plan data for dynamic optimization of the learning process to the education effect evaluation module; The educational effect evaluation module is used for educational effect evaluation, and through the educational effect evaluation, comprehensive reference data for educational effect evaluation is obtained.

Images