A smart classroom feedback analysis method, device and storage medium

Abstract

The present invention discloses a smart classroom feedback analysis method, device and storage medium. The method uses a generative adversarial network algorithm to train a feedback generation and optimization model for the smart classroom using data information related to the classroom's course theme. The feedback generation and optimization model is used to determine student feedback issues related to the course content, analyze and evaluate student feedback data submitted by students through terminals based on a machine learning algorithm, generate an analysis report, and push it to the terminals of students' parents to collect parent feedback suggestions. Based on the student feedback data and parent feedback suggestions, the feedback generation and optimization model of the smart classroom is dynamically adjusted to continuously optimize and manage the smart classroom feedback. The present invention uses a generative adversarial network (GAN) algorithm to generate and optimize student feedback issues. At the same time, based on student feedback and parent suggestions, a comprehensive feedback system is formed, which can dynamically adjust the feedback generation and optimization model to achieve continuous improvement and optimization of teaching effects.

Description

Technical Field

[0001] The present invention belongs to the field of data processing technology, and in particular relates to an intelligent classroom feedback analysis method, device and storage medium. Background Art

[0002] The traditional classroom feedback mechanism uses paper questionnaires to collect feedback from students and parents. This method is inefficient, data processing is cumbersome, and feedback is not timely enough. In addition, traditional course evaluations are usually conducted at the end of the semester and cannot reflect the effectiveness of classroom teaching and student needs in real time. After receiving feedback, teachers usually need a long time to analyze and process it, resulting in the inability to quickly implement improvement measures.

[0003] Furthermore, as the number of students increases, the efficiency and accuracy of traditional methods decline significantly, making it difficult to effectively manage and analyze large-scale feedback data. Furthermore, traditional questionnaire design is not scientific, with vague or highly leading questions, resulting in incomplete or inaccurate feedback. They often use a single template that cannot be adjusted to the specific needs of different courses or students, resulting in a lack of targeted feedback.

[0004] As teaching content and student needs evolve, traditional feedback mechanisms struggle to adjust and optimize quickly, resulting in a decline in the timeliness and effectiveness of feedback. Traditional classroom feedback mechanisms have significant shortcomings in terms of timeliness, data processing, feedback quality, personalization, parent-school interaction, and continuous optimization, limiting their effectiveness and adaptability in modern educational environments. Summary of the Invention

[0005] In response to the above-mentioned defects in the prior art, the present invention provides a smart classroom feedback analysis method, device and storage medium, the method comprising the following steps:

[0006] Step 1: Obtain data information related to the course topic of the class;

[0007] Step 2: Based on a generative adversarial network algorithm, the data information is used to train a feedback generation and optimization model of the smart classroom, where the feedback generation and optimization model is used to determine student feedback issues related to the course content;

[0008] Step 3: The student submits student feedback data based on the feedback question through the terminal, and the student feedback data is analyzed and evaluated based on the machine learning algorithm to generate an analysis report;

[0009] Step 4: Push the analysis report to the terminal of the student's parents and collect parent feedback suggestions. Based on the student feedback data and the parent feedback suggestions, dynamically adjust the feedback generation and optimization model of the smart classroom;

[0010] Step 5: Continuously execute steps 2 to 4 to achieve continuous optimization and management of smart classroom feedback.

[0011] The data information related to the course theme is collected in the classroom, including course content data, student performance data, classroom interaction data and external data. The data information is represented as follows:

[0012] D={Dc,Ds,Di,De}

[0013] Among them, Dc is course content data, Ds is student performance data, Di is classroom interaction data, and De is external data.

[0014] After obtaining the data information, the method further includes extracting features of the data information based on natural language processing (NLP) and performing cleaning and preprocessing, including:

[0015] The features of data information D are extracted based on TF-IDF and expressed as:

[0016]

[0017] DF(y)=Count(d∈D|y∈d)

[0018] Where TF(y,d) is the frequency of feature y in data document d, N is the total number of documents, Count(y,d) indicates the number of times feature y appears in document d, Total() indicates the total number of features in the document, and DF(y) is the number of documents containing feature y.

[0019] The sentiment analysis algorithm is used to calculate the sentiment score, which is expressed as:

[0020]

[0021] Among them, S is the total sentiment score, w i is the feature in the document, score(w i ) Feature w i Sentiment score;

[0022] The data cleaning process is expressed as:

[0023] D clean =Clean(D)

[0024] Among them, Clean() is the data cleaning process, including removing duplicates, filling missing values ​​and standardization.

[0025] Among them, the characteristics of the data information are integrated to form a new data set, and the data is clustered based on the K-means clustering algorithm, which is expressed as:

[0026] Dfeatures ={(D clean ,TF-IDF(y,d),S)}

[0027]

[0028] Among them, x i is the data point, μ j is the jth cluster center, and k is the number of clusters.

[0029] Among them, the features and labels for training the model are extracted from the data information obtained in step 1. The features include course content, student performance and interaction data, which are expressed as:

[0030] F={f1,f2,…,f n}

[0031] Among them, f i represents the i-th feature extracted from the data;

[0032] The tag includes the content of the student's feedback based on a predefined feedback content list, which is represented as:

[0033] L={l1,l2,…,l m}

[0034] Among them, l i Indicates the i-th feedback content;

[0035] A feedback generation and optimization model is constructed based on the generative adversarial network algorithm GAN. The adversarial network algorithm GAN includes a generator and a discriminator, which are defined as:

[0036]

[0037] Among them, G(z) is the fake data generated by the generator, and D(x) is the judgment of the discriminator on the real data;

[0038] The generator generates content similar to real student feedback, which is expressed as:

[0039] G:Z→L

[0040] Where Z is the random noise input;

[0041] The discriminator distinguishes the real feedback from the generated feedback, which is expressed as:

[0042] D:L→[0,1]

[0043] Among them, the output value represents the probability that the feedback is true;

[0044] Initialize the parameters of the generator and discriminator to random values, expressed as:

[0045]

[0046] Among them, θ G is the parameter vector of the generator, θ D is the parameter vector of the discriminator, N(0,0.02) is the process of sampling from the normal distribution;

[0047] For each training epoch t, ​​the following steps are performed:

[0048] Train the discriminator and update the discriminator using the real feedback and the generated feedback, expressed as:

[0049]

[0050]

[0051] in, is the loss function of the discriminator, is the parameter of the discriminator in the tth round, is the parameter of the discriminator after updating in the t+1th round, η is the learning rate, is the loss function of the discriminator About parameter θ D The gradient of E x～pdata For the real data distribution p data The expected value of the sample x in the equation, D(x) is the output of the discriminator for the real sample x, G(z) is the sample generated by the generator, and E z～pz For the potential distribution p z The expected value of the noise z sampled in, D(G(z)) is the output of the discriminator for the generated sample G(z);

[0052] Train the generator and update the parameters of the generator, expressed as:

[0053]

[0054] in, Denote as the loss function of the generator, is the parameter of the generator in the tth round, is the parameter of the generator after the update in the t+1th round, η is the learning rate, is the loss function of the discriminator About parameter θ D The gradient of E z～pz For the potential distribution p z The expected value of the noise z sampled in, D(G(z)) is the discriminator's prediction of the sample G(z) generated by the generator;

[0055] The feedback generation and optimization model are optimized based on the adaptive learning rate method, which is expressed as:

[0056] m t =β1m t-1 +(1-β1)g t

[0057]

[0058] Among them, m t is the first-order moment estimate of the current time step t, m t-1 is the first-order moment value of the previous time step t-1, β1 is the momentum decay coefficient, g t is the gradient of the current time step t, v t is the second-order moment estimate of the current time step t, v t-1 is the second-order moment value at the previous time step t-1, β2 is the second-order moment attenuation coefficient, θ is the model parameter, including the weight and / or bias to be optimized, η is the learning rate, and ∈ is a constant;

[0059] Based on the trained feedback generation and optimization model, a list of student feedback content related to the course content is generated, expressed as:

[0060] R=G(Z)

[0061] Among them, R is the generated feedback content set.

[0062] The feedback content set includes understanding of knowledge points, classroom interaction, homework feedback, suggestions on teaching methods, emotions and attitudes.

[0063] The feedback data Fs submitted by the students is obtained and stored. The Fs contains multiple fields, which are represented as follows:

[0064] D feedback =Insert(Fs)

[0065] Fs={u,M,R,c}

[0066] Where u is the student's unique identifier ID, M is the course ID, R is the submitted feedback content, and c is the submission timestamp;

[0067] The data is cleaned and preprocessed, expressed as:

[0068] D processed =Clean(D feedback )

[0069] Among them, the Clean function includes noise removal and text standardization;

[0070] Based on natural language processing technology, features are extracted from the feedback content and expressed as:

[0071] T=ExtractFeatures(R)

[0072] Wherein, T is the extracted feature set, which includes sentiment scores, keywords, and topic models;

[0073] Calculate the sentiment score S of the feedback based on the sentiment analysis model R

[0074] S R =SentimentAnalysis(R)

[0075] The sentiment score range is [-1,1], where -1 represents negative sentiment and 1 represents positive sentiment;

[0076] Combine the processed feedback data with the student performance data to form a training set:

[0077] D train ={Fs,P}

[0078] Among them, P represents student performance data, including test scores and class participation;

[0079] Model training is based on the random forest machine learning algorithm, and the loss function is expressed as:

[0080]

[0081] Among them, y i is the actual value, is the predicted value, N is the number of samples;

[0082] The model parameters θ are trained using the back-propagation algorithm, which is expressed as:

[0083]

[0084] Among them, θ(t) is the model parameter of the current iteration t, θ(t+) is the model parameter after the update at the k+1th iteration, η is the learning rate, The gradient of the loss function;

[0085] Generate an analysis report based on the analysis results. The analysis report includes: teaching effect evaluation, knowledge point mastery, and student sentiment analysis results, expressed as:

[0086] R={E,K,S R}

[0087] Among them: E is the teaching effect evaluation, K is the mastery of knowledge points, S R The sentiment analysis results.

[0088] Among them, the push notification system is used to send the report to the parent terminal, and the parent feedback suggestions are collected and stored, which are expressed as:

[0089] F p ={parent_id, suggestions, timestamp}

[0090] D parent_feedback =Insert(F p )

[0091] Among them, F P It is a collection of parent feedback suggestions. parent_id is the parent's unique identifier, suggestions is the feedback suggestions or opinions put forward by the parent, and timestamp is the timestamp of recording the feedback submission.

[0092] Student feedback data D feedback and parent feedback data parent_feedback The combined packages form a new training set, expressed as:

[0093] D combined =D feedback ∪D parent_feedback

[0094] The feedback data is given different weights through a weighting mechanism, which can be expressed as:

[0095] W={w s ,w p}

[0096] Where: w s is the weight of student feedback, w p the weight given to parent feedback;

[0097] Dynamic adjustment is performed based on the weighted loss function, which is expressed as:

[0098]

[0099] Where: L student is the loss based on student feedback data, L parent For losses based on parent feedback data;

[0100] Based on the new training set D combined Retrain the model and update the model parameters θ, which can be expressed as:

[0101]

[0102] Among them, θ(t) is the model parameter at the current time t, θ(t+1) is the updated model parameter at the next time t+1, η is the learning rate, The gradient of the loss function;

[0103] The new feedback data and the adjusted model are used again for the next round of student feedback question generation.

[0104] The present invention obtains data information related to the course theme of the classroom, and uses the data information to train the feedback generation and optimization model of the smart classroom based on the generative adversarial network algorithm. The feedback generation and optimization model is used to determine student feedback issues related to the course content. Students submit student feedback data based on the feedback issues through the terminal. The student feedback data is analyzed and evaluated based on the machine learning algorithm, and an analysis report is generated. The analysis report is pushed to the terminal of the student's parents, and parent feedback suggestions are collected. Based on the student feedback data and the parent feedback suggestions, the feedback generation and optimization model of the smart classroom is dynamically adjusted to continuously optimize the management of smart classroom feedback. The present invention uses the generative adversarial network (GAN) algorithm to generate and optimize student feedback issues. At the same time, based on student feedback and parent suggestions, a comprehensive feedback system is formed, which can dynamically adjust the feedback generation and optimization model to achieve continuous improvement and optimization of teaching effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0105] The above and other objects, features and advantages of the exemplary embodiments of the present disclosure will become readily understood by reading the following detailed description with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present disclosure are shown in an illustrative and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0106] Figure 1 4 is a flow chart illustrating a method for analyzing smart classroom feedback according to an embodiment of the present invention. DETAILED DESCRIPTION

[0107] To make the objectives, technical solutions, and advantages of the present invention more apparent, the present invention will be further described in detail below with reference to the accompanying drawings. It is apparent that the embodiments described are only some, not all, of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort are intended to fall within the scope of protection of the present invention.

[0108] The terms used in the embodiments of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The singular forms "a," "an," "the," and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, and unless the context clearly indicates otherwise, "a plurality" generally includes at least two.

[0109] It should be understood that although the terms "first," "second," "third," etc. may be used to describe "...," these "..." should not be limited to these terms. These terms are merely used to distinguish "...." For example, "first..." could also be referred to as "second...", and similarly, "second..." could also be referred to as "first..." without departing from the scope of the present invention.

[0110] It should be understood that the term "and / or" as used herein is merely a description of the relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " in this document generally indicates that the associated objects are in an "or" relationship.

[0111] As used herein, the words "if" and "if" may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to the determination" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)," depending on the context.

[0112] It should also be noted that the terms "include," "comprises," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a product or device comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such product or device. In the absence of further limitations, an element defined by the phrase "comprises a..." does not exclude the presence of other identical elements in the product or device comprising the element.

[0113] Traditional classroom feedback mechanisms have significant shortcomings in terms of feedback timeliness, data processing, feedback quality, personalization, home-school interaction, and continuous optimization, limiting their effectiveness and adaptability in modern educational environments. Therefore, an effective method is needed to address the challenges of existing classroom feedback mechanisms by improving the quality and efficiency of feedback, fostering home-school collaboration, and providing more effective support for classroom teaching improvements.

[0114] like Figure 1 As shown, the present invention discloses a smart classroom feedback analysis method, the method comprising:

[0115] Step 1: Obtain data information related to the course topic of the class.

[0116] By acquiring data related to the course topic, we establish a reliable data foundation, providing the necessary information support for subsequent feedback generation and analysis. The collected data will serve as the basis for generating student feedback questions related to the course content, ensuring that the generated questions are closely related to the actual teaching content. By analyzing data related to the course topic, we can identify difficulties and doubts that students may encounter during the learning process, thereby customizing feedback questions and improving the relevance and effectiveness of feedback.

[0117] This data provides essential input for training the feedback generation and optimization model based on the Generative Adversarial Network, ensuring that the model can better understand and generate relevant feedback. As the course progresses, real-time data related to the class topic is acquired, enabling dynamic monitoring of student learning progress, identifying issues and making adjustments promptly. By acquiring data related to the class topic, the accuracy and pertinence of subsequent feedback generation and analysis can be ensured, promoting improved and continuous optimization of teaching effectiveness.

[0118] In one embodiment, data information related to the course topic is collected in class, including course content data, student performance data, classroom interaction data, and external data. The data information is represented as follows:

[0119] D={Dc,Ds,Di,De}

[0120] Among them, Dc is course content data, Ds is student performance data, Di is classroom interaction data, and De is external data.

[0121] By collecting different types of data, teachers can gain a comprehensive understanding of all aspects of the classroom, thereby better understanding students' learning status and needs. This diverse data provides rich information support, enabling teachers to make more scientific teaching decisions based on data and optimize teaching strategies. Analyzing student performance data and classroom interaction data can identify individual differences among students, thereby providing personalized learning advice and assistance. The collection of external data not only helps teachers understand students' learning progress but also provides parents with detailed information on their children's performance in school, promoting communication and cooperation between home and school.

[0122] Combining course content, student performance, classroom interactions, and external data allows for a comprehensive analysis of student learning from multiple dimensions, enhancing both the depth and breadth of analysis. The integration of multiple data sources enables more accurate and specific feedback questions, tailored to students' specific needs. By collecting course content data, student performance data, classroom interaction data, and external data, teachers gain a comprehensive understanding of classroom situations, enabling them to make more informed teaching decisions, provide personalized learning support, promote parent-school collaboration, and ultimately achieve improved and continuously optimized teaching outcomes.

[0123] In one embodiment, after obtaining the data information, the method further includes extracting features of the data information based on natural language processing (NLP) and performing cleaning and preprocessing, including:

[0124] The features of data information D are extracted based on TF-IDF and expressed as:

[0125]

[0126] DF(y)=Count(d∈D|y∈d)

[0127] Where TF(y,d) is the frequency of feature y in data document d, N is the total number of documents, Count(y,d) indicates the number of times feature y appears in document d, Total() indicates the total number of features in the document, and DF(y) is the number of documents containing feature y.

[0128] The sentiment analysis algorithm is used to calculate the sentiment score, which is expressed as:

[0129]

[0130] Among them, S is the total sentiment score, w i is the feature in the document, score(w i ) Feature w i Sentiment score;

[0131] The data cleaning process is expressed as:

[0132] D clean =Clean(D)

[0133] Among them, Clean() is the data cleaning process, including removing duplicates, filling missing values ​​and standardization.

[0134] Feature extraction and data cleaning improve the quality of input data and ensure the accuracy of subsequent analysis and modeling. Features extracted using TF-IDF make the data more interpretable. Sentiment analysis, by quantifying sentiment scores, helps teachers understand students' emotional states and attitudes, enabling them to better adjust teaching strategies. This step, through NLP-based feature extraction, sentiment analysis, and data cleaning, provides strong support for data processing and analysis, ensuring that subsequent feedback generation and teaching decisions are more scientific, accurate, and efficient.

[0135] In one embodiment, the features of the data information are integrated to form a new data set, and the data is clustered based on the K-means clustering algorithm, which is expressed as:

[0136] D features ={(D clean,TF-IDF(y,d),S)}

[0137]

[0138] Among them, x i is the data point, μ j is the jth cluster center, and k is the number of clusters.

[0139] The cleaned data, TF-IDF features, and sentiment scores are integrated to form a comprehensive dataset, providing comprehensive information support for subsequent analysis. K-means clustering can help identify underlying patterns and structures in the data, allowing teachers to understand the distribution of student performance and feedback. Through clustering, teachers can identify different types of student groups, such as high-performing and low-performing students, and implement targeted interventions.

[0140] By integrating data information features and applying the K-means clustering algorithm, we can effectively identify the characteristics and behavior patterns of student groups, provide data support for teaching decisions, and improve teaching effectiveness and student learning experience.

[0141] Step 2: Based on a generative adversarial network algorithm, the data information is used to train a feedback generation and optimization model of the smart classroom, where the feedback generation and optimization model is used to determine student feedback issues related to the course content.

[0142] Through GAN adversarial training, the generated feedback questions will be closer to the real situation and can effectively reflect students' needs and course content. The feedback generation and optimization model can continuously adjust the content of the generated questions based on real-time data and student feedback to ensure that it continues to adapt to course changes.

[0143] By using a generative adversarial network algorithm to train a feedback generation and optimization model, we can effectively generate student feedback questions that are relevant to the course content. This process not only improves the quality and personalization of feedback, but also reduces the workload of teachers, enhances the flexibility and adaptability of teaching, and thus improves the overall teaching effect.

[0144] In one embodiment, features and labels for training the model are extracted from the data information obtained in step 1. The features include course content, student performance, and interaction data, and are expressed as:

[0145] F={f1,f2,…,f n}

[0146] Among them, f i represents the i-th feature extracted from the data;

[0147] The tag includes the content of the student's feedback based on a predefined feedback content list, which is represented as:

[0148] L={l1,l2,…,l m}

[0149] Among them, l i Indicates the i-th feedback content;

[0150] A feedback generation and optimization model is constructed based on the generative adversarial network algorithm GAN. The adversarial network algorithm GAN includes a generator and a discriminator, which are defined as:

[0151]

[0152] Among them, G(z) is the fake data generated by the generator, and D(x) is the judgment of the discriminator on the real data;

[0153] The generator generates content similar to real student feedback, which is expressed as:

[0154] G:Z→L

[0155] Where Z is the random noise input;

[0156] The discriminator distinguishes the real feedback from the generated feedback, which is expressed as:

[0157] D:L→[0,1]

[0158] Among them, the output value represents the probability that the feedback is true;

[0159] Initialize the parameters of the generator and discriminator to random values, expressed as:

[0160]

[0161] Among them, θ G is the parameter vector of the generator, θ D is the parameter vector of the discriminator, N(0,0.02) is the process of sampling from the normal distribution;

[0162] For each training epoch t, ​​the following steps are performed:

[0163] Train the discriminator and update the discriminator using the real feedback and the generated feedback, expressed as:

[0164]

[0165] in, is the loss function of the discriminator, is the parameter of the discriminator in the tth round, is the parameter of the discriminator after updating in the t+1th round, η is the learning rate, is the loss function of the discriminator About parameter θ D The gradient of E x～pdata For the real data distribution p data The expected value of the sample x in the equation, D(x) is the output of the discriminator for the real sample x, G(z) is the sample generated by the generator, and E z～pz For the potential distribution p z The expected value of the noise z sampled in, D(G(z)) is the output of the discriminator for the generated sample G(z);

[0166] Train the generator and update the parameters of the generator, expressed as:

[0167]

[0168] in, Denote as the loss function of the generator, is the parameter of the generator in the tth round, is the parameter of the generator after the update in the t+1th round, η is the learning rate, is the loss function of the discriminator About parameter θ D The gradient of E z～pz For the potential distribution p z The expected value of the noise z sampled in, D(G(z)) is the discriminator's prediction of the sample G(z) generated by the generator;

[0169] The feedback generation and optimization model are optimized based on the adaptive learning rate method, which is expressed as:

[0170] m t =β1m t-1 +(1-β1)g t

[0171]

[0172] Among them, m t is the first-order moment estimate of the current time step t, m t-1 is the first-order moment value of the previous time step t-1, β1 is the momentum decay coefficient, g t is the gradient of the current time step t, v t is the second-order moment estimate of the current time step t, v t-1 is the second-order moment value at the previous time step t-1, β2 is the second-order moment attenuation coefficient, θ is the model parameter, including the weight and / or bias to be optimized, η is the learning rate, and ∈ is a constant;

[0173] Based on the trained feedback generation and optimization model, a list of student feedback content related to the course content is generated, expressed as:

[0174] R=G(Z)

[0175] Among them, R is the generated feedback content set.

[0176] In one embodiment, the feedback content set includes understanding of knowledge points, classroom interaction, homework feedback, suggestions on teaching methods, emotions and attitudes.

[0177] Step 3: The student submits student feedback data based on the feedback question through the terminal, and the student feedback data is analyzed and evaluated based on the machine learning algorithm, and an analysis report is generated.

[0178] Students submit feedback data in response to generated questions through a terminal (such as a mobile app, website, or learning management system). This feedback can be in the form of text, multiple-choice answers, or scores. Students can submit feedback instantly, forming a dynamic feedback mechanism that allows teachers to promptly understand students' needs and course issues.

[0179] Machine learning algorithms are used to analyze collected feedback data. These include: Classification Analysis: Identifying different types of feedback (such as positive, negative, and neutral) to help understand students' emotional state; Cluster Analysis: Aggregating similar feedback data to identify potential common issues or needs; and Trend Analysis: Monitoring feedback data trends to assess teaching effectiveness and student satisfaction. By analyzing feedback data through machine learning, teachers can gain data-driven insights and make more informed teaching decisions.

[0180] Based on the analysis results, a comprehensive report is generated, including an overview of student feedback, identification of key issues and themes, suggestions for improvement, and action plans. By analyzing student feedback data using machine learning algorithms, the smart classroom provides a real-time feedback mechanism and data-driven decision support. This process not only improves the efficiency and accuracy of feedback but also enhances student engagement, promotes continuous improvement in teaching, and ultimately enhances teaching effectiveness and student satisfaction.

[0181] In one embodiment, feedback data Fs submitted by students is obtained and stored. Fs includes multiple fields, which are represented as follows:

[0182] D feedback =Insert(Fs)

[0183] Fs={u,M,R,c}

[0184] Where u is the student's unique identifier ID, M is the course ID, R is the submitted feedback content, and c is the submission timestamp;

[0185] The data is cleaned and preprocessed, expressed as:

[0186] Dprocessed =Clean(D feedback )

[0187] Among them, the Clean function includes noise removal and text standardization;

[0188] Based on natural language processing technology, features are extracted from the feedback content and expressed as:

[0189] T=ExtractFeatures(R)

[0190] Wherein, T is the extracted feature set, which includes sentiment scores, keywords, and topic models;

[0191] Calculate the sentiment score S of the feedback based on the sentiment analysis model R

[0192] S R =SentimentAnalysis(R)

[0193] The sentiment score range is [-1,1], where -1 represents negative sentiment and 1 represents positive sentiment;

[0194] Combine the processed feedback data with the student performance data to form a training set:

[0195] D train ={Fs,P}

[0196] Among them, P represents student performance data, including test scores and class participation;

[0197] Model training is based on the random forest machine learning algorithm, and the loss function is expressed as:

[0198]

[0199] Among them, y i is the actual value, is the predicted value, N is the number of samples;

[0200] The model parameters θ are trained using the back-propagation algorithm, which is expressed as:

[0201]

[0202] Among them, θ(t) is the model parameter of the current iteration t, θ(t+) is the model parameter after the update at the k+1th iteration, η is the learning rate, The gradient of the loss function;

[0203] Generate an analysis report based on the analysis results. The analysis report includes: teaching effect evaluation, knowledge point mastery, and student sentiment analysis results, expressed as:

[0204] R={E,K,S R}

[0205] Among them: E is the teaching effect evaluation, K is the mastery of knowledge points, S R Sentiment analysis results.

[0206] Step 4: Push the analysis report to the terminal of the student's parents and collect parent feedback suggestions. Based on the student feedback data and the parent feedback suggestions, dynamically adjust the feedback generation and optimization model of the smart classroom.

[0207] The generated analysis report is delivered to parents' devices via apps, email, or other digital channels, ensuring they have timely access to information relevant to their children's learning. Parents can provide feedback on the analysis report, offering suggestions and opinions. This feedback can include opinions on course content, teaching methods, or other related matters.

[0208] Collected parent feedback suggestions are combined with student feedback data for comprehensive analysis to identify parental perceptions of instruction and their correlation with student feedback. Based on the results of this comprehensive analysis, the smart classroom's feedback generation and optimization model is dynamically adjusted to better meet the needs of students and parents. This may include modifying generated feedback questions, adjusting teaching strategies, or optimizing course content.

[0209] By distributing analysis reports and collecting parent feedback, communication between home and school is strengthened, fostering a shared focus on student learning and growth. Parent feedback provides teachers with insights into student performance and attitudes in the home environment, complementing in-classroom student feedback and providing a more comprehensive feedback perspective. Distributing analysis reports to parents and collecting their feedback enhances communication and collaboration between home and school. This process not only provides a more comprehensive feedback perspective but also fosters family engagement, dynamically adjusting the smart classroom's feedback generation and optimization model, ultimately improving personalized education and satisfaction.

[0210] In one embodiment, a push notification system is used to send a report to a parent terminal, and parent feedback suggestions are collected and stored, which are expressed as:

[0211] F p ={parent_id, suggestions, timestamp}

[0212] D parent_feedback =Insert(F p )

[0213] Among them, F PIt is a collection of parent feedback suggestions. parent_id is the parent's unique identifier, suggestions is the feedback suggestions or opinions put forward by the parent, and timestamp is the timestamp of recording the feedback submission.

[0214] Student feedback data D feedback and parent feedback data parent_feedback The combined packages form a new training set, expressed as:

[0215] D combined =D feedback ∪D parent_feedback

[0216] The feedback data is given different weights through a weighting mechanism, which can be expressed as:

[0217] W={w s ,w p}

[0218] Where: w s is the weight of student feedback, w p the weight given to parent feedback;

[0219] Dynamic adjustment is performed based on the weighted loss function, which is expressed as:

[0220]

[0221] Where: L student is the loss based on student feedback data, L parent For losses based on parent feedback data;

[0222] Based on the new training set D combined Retrain the model and update the model parameters θ, which can be expressed as:

[0223]

[0224] Among them, θ(t) is the model parameter at the current time t, θ(t+1) is the updated model parameter at the next time t+1, η is the learning rate, The gradient of the loss function;

[0225] The new feedback data and the adjusted model are used again for the next round of student feedback question generation.

[0226] Step 5: Continuously execute steps 2 to 4 to achieve continuous optimization and management of smart classroom feedback.

[0227] The present invention obtains data information related to the course theme of the classroom, and uses the data information to train the feedback generation and optimization model of the smart classroom based on the generative adversarial network algorithm. The feedback generation and optimization model is used to determine student feedback issues related to the course content. Students submit student feedback data based on the feedback issues through the terminal. The student feedback data is analyzed and evaluated based on the machine learning algorithm, and an analysis report is generated. The analysis report is pushed to the terminal of the student's parents, and parent feedback suggestions are collected. Based on the student feedback data and the parent feedback suggestions, the feedback generation and optimization model of the smart classroom is dynamically adjusted to continuously optimize the management of smart classroom feedback. The present invention uses the generative adversarial network (GAN) algorithm to generate and optimize student feedback issues. At the same time, based on student feedback and parent suggestions, a comprehensive feedback system is formed, which can dynamically adjust the feedback generation and optimization model to achieve continuous improvement and optimization of teaching effects.

[0228] It should be noted that the computer-readable medium mentioned above in the present disclosure may be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or component, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, device, or component. In the present disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium may be transmitted using any suitable medium, including but not limited to wires, optical cables, RF (radio frequency), etc., or any suitable combination thereof.

[0229] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0230] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on the remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0231] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the module, program segment, or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of the boxes in the block diagram and / or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0232] The units involved in the embodiments described in this disclosure may be implemented in software or hardware, wherein the name of a unit does not necessarily limit the unit itself.

[0233] The above introduces the preferred embodiments of the present invention, which is intended to make the spirit of the present invention clearer and easier to understand, and is not intended to limit the present invention. Any modifications, replacements, and improvements made within the spirit and principles of the present invention should be included in the scope of protection outlined by the claims attached to the present invention.

Claims

1. A method for analyzing smart classroom feedback, comprising the following steps: Step 1: Obtain data information related to the course topic of the class; Step 2: Based on a generative adversarial network algorithm, the data information is used to train a feedback generation and optimization model of the smart classroom, where the feedback generation and optimization model is used to determine student feedback issues related to the course content; Step 3: The student submits student feedback data based on the feedback question through the terminal, and the student feedback data is analyzed and evaluated based on the machine learning algorithm to generate an analysis report; Step 4: Push the analysis report to the terminal of the student's parents and collect parent feedback suggestions. Based on the student feedback data and the parent feedback suggestions, dynamically adjust the feedback generation and optimization model of the smart classroom; Step 5: Continuously execute steps 2 to 4 to achieve continuous optimization and management of smart classroom feedback; Also includes, Extract features and labels for model training from the data information obtained in step 1. The features include course content, student performance, and interaction data, which are expressed as: F={f1,f2,…,f n } Among them, f i represents the i-th feature extracted from the data; The tag includes the content of the student's feedback based on a predefined feedback content list, which is represented as: L＝{l1,l2,…,l m} Among them, l i Indicates the i-th feedback content; A feedback generation and optimization model is constructed based on the generative adversarial network algorithm GAN. The adversarial network algorithm GAN includes a generator and a discriminator, which are defined as: Among them, G(z) is the fake data generated by the generator, and D(x) is the judgment of the discriminator on the real data; The generator generates content similar to real student feedback, which is expressed as: G:Z→L Where Z is the random noise input; The discriminator distinguishes the real feedback from the generated feedback, which is expressed as: D:L→[0,1] Among them, the output value represents the probability that the feedback is true; Initialize the parameters of the generator and discriminator to random values, expressed as: Among them, θ G is the parameter vector of the generator, θ D is the parameter vector of the discriminator, N(0,0.02) is the process of sampling from the normal distribution; For each training epoch t, ​​the following steps are performed: Train the discriminator and update the discriminator using the real feedback and the generated feedback, expressed as: in, is the loss function of the discriminator, is the parameter of the discriminator in the tth round, is the parameter of the discriminator after updating in the t+1th round, η is the learning rate, is the loss function of the discriminator About parameter θ D The gradient of E x～pdata For the real data distribution p data The expected value of the sample x in the equation, D(x) is the output of the discriminator for the real sample x, G(z) is the sample generated by the generator, and E z～pz For the potential distribution p z The expected value of the noise z sampled in, D(G(z)) is the output of the discriminator for the generated sample G(z); Train the generator and update the parameters of the generator, expressed as: in, Denote as the loss function of the generator, is the parameter of the generator in the tth round, is the parameter of the generator after the update in the t+1th round, η is the learning rate, is the loss function of the generator About parameter θ G The gradient of E z～pz For the potential distribution p z The expected value of the noise z sampled in, D(G(z)) is the discriminator's prediction of the sample G(z) generated by the generator; The feedback generation and optimization model are optimized based on the adaptive learning rate method, which is expressed as: m t =β1m t-1 +(1-β1)g t Among them, m t is the first-order moment estimate of the current time step t, m t-1 is the first-order moment value of the previous time step t-1, β1 is the momentum decay coefficient, g t is the gradient of the current time step t, v t is the second-order moment estimate of the current time step t, v t-1 is the second-order moment value of the previous time step t-1, β2 is the second-order moment attenuation coefficient, θ t is the parameter of the model at the current time step t, θ t-1 are the parameters of the model at the previous time step t-1, including the weights and / or biases that need to be optimized, η is the learning rate, and ∈ is a constant; Based on the trained feedback generation and optimization model, a list of student feedback content related to the course content is generated, expressed as: R=G(Z) Where R is the generated feedback content set, which includes understanding of knowledge points, classroom interaction, homework feedback, suggestions on teaching methods, emotions and attitudes; Also includes, Obtain and store the feedback data Fs submitted by the student. The Fs contains multiple fields, represented as follows: D feedback =Insert(Fs) Fs={u,M,R,c} Where u is the student's unique identifier ID, M is the course ID, R is the submitted feedback content, and c is the submission timestamp; The data is cleaned and preprocessed, expressed as: D processed =Clean(D feedback ) Among them, the Clean function includes noise removal and text standardization; Based on natural language processing technology, features are extracted from the feedback content and expressed as: T=ExtractFeatures(R) Wherein, T is the extracted feature set, which includes sentiment scores, keywords, and topic models; Calculate the sentiment score S of the feedback based on the sentiment analysis model R S R =SentimentAnalysis(R) The sentiment score range is [-1,1], where -1 represents negative sentiment and 1 represents positive sentiment; Combine the processed feedback data with the student performance data to form a training set: D train ={Fs,P} Among them, P represents student performance data, including test scores and class participation; Model training is based on the random forest machine learning algorithm, and the loss function is expressed as: Among them, y i is the actual value, is the predicted value, N is the number of samples; The model parameters θ are trained using the back-propagation algorithm, which is expressed as: Where θ(t) is the model parameter of the current iteration t, θ(t+1) is the updated model parameter of the next iteration t+1, η is the learning rate, is the loss function gradient; Generate an analysis report based on the analysis results. The analysis report includes: teaching effect evaluation, knowledge point mastery, and student sentiment analysis results, expressed as: R={E,K,S R } Among them: E is the teaching effect evaluation, K is the mastery of knowledge points, S R Sentiment analysis results.

2. The intelligent classroom feedback analysis method according to claim 1, characterized in that: Collect data information related to the course topic in class, including course content data, student performance data, classroom interaction data and external data. The data information is represented as follows: D={Dc,Ds,Di,De} Among them, Dc is course content data, Ds is student performance data, Di is classroom interaction data, and De is external data.

3. The intelligent classroom feedback analysis method according to claim 2, characterized in that: After obtaining the data information, the method further includes extracting features of the data information based on natural language processing (NLP) and performing cleaning and preprocessing, including: The features of data information D are extracted based on TF-IDF and expressed as: DF(y)=Count(d∈D|y∈d) Where TF(y,d) is the frequency of feature y in data document d, N is the total number of documents, Count(y,d) indicates the number of times feature y appears in document d, Total() indicates the total number of features in the document, and DF(y) is the number of documents containing feature y. The sentiment analysis algorithm is used to calculate the sentiment score, which is expressed as: Among them, S is the total sentiment score, w i is the feature in the document, score(w i ) Feature w i Sentiment score; The data cleaning process is expressed as: D clean =Clean(D) Among them, Clean() is the data cleaning process, including removing duplicates, filling missing values ​​and standardization.

4. The intelligent classroom feedback analysis method according to claim 3, characterized in that: The features of the data information are integrated to form a new data set, and the data is clustered based on the K-means clustering algorithm, which is expressed as: D features ={(D clean ,TF-IDF(y,d),S)} Among them, x i is the data point, μ j is the jth cluster center, and k is the number of clusters.

5. The intelligent classroom feedback analysis method according to claim 4, characterized in that: Use the push notification system to send reports to parents’ terminals, collect and store parent feedback suggestions, expressed as: F p ={parent_id,suggestions,timestamp} D parent_feedback =Insert(F p ) Among them, F P It is a collection of parent feedback suggestions. parent_id is the parent's unique identifier, suggestions is the feedback suggestions or opinions put forward by the parent, and timestamp is the timestamp of recording the feedback submission. Student feedback data D feedback and parent feedback data parent_feedback The combined packages form a new training set, expressed as: D combined =D feedback ∪D parent_feedback The feedback data is given different weights through a weighting mechanism, which can be expressed as: W={w s ,In p } Where: w s is the weight of student feedback, w p the weight given to parent feedback; Dynamic adjustment is performed based on the weighted loss function, which is expressed as: Where: L student is the loss based on student feedback data, L parent For losses based on parent feedback data; Based on the new training set D combined Retrain the model and update the model parameters θ, which can be expressed as: Among them, θ(t) is the model parameter at the current time t, θ(t+1) is the updated model parameter at the next time t+1, η is the learning rate, is the loss function gradient; The new feedback data and the adjusted model are used again for the next round of student feedback question generation.

6. An intelligent classroom feedback analysis device, comprising: at least one processor; as well as at least one memory including computer program code, The at least one memory and the computer program code are configured to, together with the at least one processor, enable the apparatus to perform the method according to any one of claims 1 to 5.

7. A computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the method according to any one of claims 1 to 5 when executed by a processor.

Images