Method and device for predicting online open course learner satisfaction and electronic equipment

Abstract

The invention relates to a method and device for predicting online open course learner satisfaction and electronic equipment, and the method comprises the steps: predicting the online open course satisfaction of students through an MLP and RBF neural network model by using a virtual learning environment of a large-scale online teaching and learning platform and combining learning behavior data in a learning management system (LMS); the model comprises a data acquisition and processing module, a multilayer perceptron (MLP) module, a radial basis function (RBF) neural network module, a classification tree module and a control block, the data acquisition and processing module is used for generating a training and testing data set, and the MLP and RBF neural network model predicts the satisfaction degree of a learner. The MLP model carries out feature extraction through a multi-layer perceptron structure and different activation functions, the RBF model measures the distance between input data and a center by using a radial basis function to realize feature extraction, the classification tree is used for judging a prediction model to which a data point belongs, the control block integrates features from the MLP and RBF neural network models, and the RBF model is used for determining a prediction model to which the data point belongs. Experimental results show that the prediction accuracy of low-satisfaction-degree learners and high-satisfaction-degree learners can be improved at the same time through the combination scheme of the MLP and the RBF, the method can be applied to learner satisfaction degree prediction of various online open courses, an educational institution is helped to know the satisfaction degree condition of students in time, course design and teaching strategies are optimized, and the teaching efficiency is improved. And important support is provided for teaching reform and optimization in the field of online education.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of computer science and educational technology, and specifically relates to a method, device and electronic equipment for predicting online open course learner satisfaction. BACKGROUND

[0002] Currently, with the popularity of online open courses, more and more learners participate in these courses to acquire knowledge and skills. However, course providers and educational institutions face an important challenge, which is how to effectively evaluate learners' satisfaction and learning experience of the course in order to further optimize course design and teaching strategies. Traditional evaluation methods often rely on learners' feedback surveys or regular test results, which are limited by time and resources, and may not provide real-time and personalized feedback.

[0003] Therefore, it is particularly important and necessary to develop a method, device and electronic equipment for predicting online open course learner satisfaction. However, the traditional learner satisfaction evaluation method has the following problems: (1) relying on static data: traditional methods often rely on end-of-semester surveys or test results, which cannot reflect learners' real feelings in real time. (2) Lack of personalization: most methods are based on overall course evaluation, lacking consideration of individual learners' personalized needs. (3) Time delay: traditional evaluation methods usually take time to collect and process data, which cannot respond and adjust in time. SUMMARY

[0004] The present application relates to the field of computer science and educational technology, particularly in the application of data analysis and prediction of online open courses (MOOCs) and learning management systems (LMS). The present application first uses a data acquisition and processing module to obtain experimental samples, then adopts a parallel way of MLP and RBF, respectively trains and extracts features independently, and uses a classification tree to divide the samples, finally uses a control block to predict online open course learner satisfaction.

[0005] Firstly, the online learning behavior data of learners is obtained from the learning management system (LMS), including the page browsing history H i , course progress P i , online test results R i , learning duration T i , and the number of course interactions I iThe acquired data is preprocessed, including outlier processing, missing value filling and feature normalization. The satisfaction label is obtained through a designed questionnaire, including multiple-choice questions and scoring questions. The questions focus on course content, teaching quality and overall satisfaction. The questionnaire is distributed through three channels: email, course platform announcement and social media. The collected feedback data is organized into a dataset. This time, 300 learner online learning behavior data were collected. The Likert scale 5-point option was used to evaluate the satisfaction of the respondents, where 1 represents very dissatisfied, 3 represents a neutral attitude, and 5 represents strong satisfaction. In the experiment, the 5-point label is set to 1, indicating satisfaction, and the 1-4 point label is set to 0, indicating dissatisfaction. The training set and test set are allocated in a ratio of 8:2.

[0006] The multilayer perceptron module receives and analyzes the learning behavior data of students and performs deep learning through multiple hidden layers h=5 to extract complex patterns in the learning behavior data of learners and generate satisfaction prediction based on learner behavior. The features of the data processing module are used as input. The model includes an input layer, multiple hidden layers and an output layer. The number of output layer nodes is 2, representing the probability of satisfaction and dissatisfaction. The process of each hidden layer of MLP is as follows:

[0007]

[0008] The radial basis function neural network module is configured to use Gaussian radial basis functions to measure the distance d(x i ,c j ) between the input data x i and the preset center c j , which is used to measure the distance between the learning behavior data and the preset center. The model includes an input layer, a radial basis function layer and an output layer. The RBF neural network model uses Gaussian radial basis functions as basic calculation units to measure the distance d(x i ,c j ) between the input data x i and the preset center c j . The formula is as follows:

[0009]

[0010] where σ j is the width parameter of the radial basis function. The clustering algorithm (such as K-Means clustering) is used to determine the center c j of the RBF neural network model to optimize the performance and prediction accuracy of the radial basis function.

[0011] The output formula of the RBF model is as follows:

[0012]

[0013] where β j is the weight of the j-th radial basis function, and φ is the radial basis function.

[0014] The classification tree module uses the output features of the MLP and RBF to split the data set, so that the subsets after splitting are as pure as possible in terms of labels, i.e. the data points in each subset belong to the same class as much as possible, and the splitting criteria used include information gain and Gini coefficient, information gain calculates the difference in information entropy after splitting by a certain feature, and the feature with the largest information entropy difference is selected, and Gini coefficient calculates the Gini impurity after splitting by a certain feature, and the feature with the smallest Gini impurity is selected.

[0015] MLP and RBF each have their own loss function, and the final total loss can be calculated by combining first and then backpropagating or backpropagating first and then combining, and the loss uses mean square error (MSE) or cross-entropy loss denoted as L(·,·),

[0016] The calculation of the backpropagation first and then combine method is as follows:

[0017] (1) Calculate the loss of the MLP model:

[0018]

[0019] (2) Calculate the loss of the RBF model:

[0020]

[0021] (3) Calculate the total loss:

[0022] Total Loss = α·Loss MLP +(1-α)·Loss RBF

[0023] where α is the weight, which can be set according to the performance of the model or prior knowledge, and the loss expression of this parallel method can ensure that the loss of both models affects the total loss, so that both models are optimized during the training process,

[0024] The first combination and then backpropagation method performs result fusion after model prediction, and calculates the total loss based on the fused prediction results, first calculate the fused prediction features as follows:

[0025]

[0026] The total loss is represented as:

[0027]

[0028] This invention, by combining deep learning and neural network technologies, provides a method, device, and electronic device for predicting learner satisfaction in online open courses (MOOCs). It achieves real-time performance: real-time collection and analysis of learners' online learning behavior data to quickly generate predictions of learner satisfaction; personalization: through deep learning and neural network models, personalized analysis and prediction of each learner's behavioral patterns are possible; accuracy: utilizing multilayer perceptron (MLP) and radial basis function (RBF) neural network models improves the accuracy and precision of learner satisfaction prediction; and automation: the system can automatically process and analyze large amounts of learning behavior data, reducing the workload of manual intervention and data processing. This method not only improves the accuracy and efficiency of assessment but also effectively optimizes course design and teaching strategies, thereby improving learners' overall learning experience and satisfaction. This invention can be widely applied in fields such as online open courses (MOOCs), distance education, and corporate training, providing educators and learners with personalized and optimized learning experiences. Attached Figure Description

[0029] Figure 1 This is a flowchart of the method of the present invention;

[0030] Figure 2 This is a flowchart of the data acquisition process in this invention;

[0031] Figure 3 This is the MLP network diagram in this invention;

[0032] Figure 4 This is the RBF network diagram in this invention. Detailed Implementation Plan

[0033] The present invention discloses a method, apparatus, and electronic device for predicting learner satisfaction in online open courses, comprising a network framework consisting of a data acquisition and processing module, a multilayer perceptron (MLP) module, a radial basis function (RBF) neural network module, a classification tree module, and a control block. The present invention will be described in detail below with reference to the accompanying drawings.

[0034] 1. Data acquisition and processing:

[0035] Obtain learners' online learning behavior data from the Learning Management System (LMS), including learners' page browsing history. i Course progress P i Online test results R i Learning duration T i Number of course interactions i The acquired data is preprocessed, including outlier handling, missing value imputation, and feature normalization, to ensure data quality and the effectiveness of model training.

[0036] The satisfaction label is obtained by designing a questionnaire, including multiple-choice questions and scoring questions. The questions are around course content, teaching quality and overall satisfaction. The questionnaire is distributed through email, course platform announcement and social media. The collected feedback data is organized into a dataset. This time, 300 learner online learning behavior data were collected. The Likert scale 5-point option was used to evaluate the satisfaction of the respondents, with 1 indicating very dissatisfied, 3 indicating a neutral attitude, and 5 indicating strong satisfaction. In the experiment, the 5-point label was set to 1, indicating satisfaction, and the 1-4 point label was set to 0, indicating dissatisfaction. The training set and test set were allocated in a ratio of 8:2.

[0037] Data preprocessing includes:

[0038] (1) Missing value processing: deleting records containing a large number of missing values, and using mean or interpolation method to fill in missing values for records with few missing values, leaving 246 samples;

[0039] (2) Abnormal value processing: using Z-Score or IQR method to detect and process abnormal values in the data;

[0040] (3) Encoding categorical data: converting categorical variables to numerical form, this experiment converts to One-Hot Encoding;

[0041] (4) Standardization and normalization: standardizing or normalizing numerical data to have the same scale.

[0042] 2. Multi-layer perceptron (MLP) module:

[0043] The number of input layer nodes of the MLP module is the same as the number of input features. This module designs 5 hidden layers, with the number of nodes in each hidden layer set to half of the input. The Dropout layer is used to randomly discard part of the units to prevent model overfitting. The number of output layer nodes is equal to the number of satisfaction categories, which is 2.

[0044] The hidden layer uses the ReLU activation function, and the output layer uses the Sigmoid function to realize the binary classification problem. The Adam optimizer is used with an initial learning rate of 0.001, a batch size of 16, and a maximum number of iterations of 50.

[0045] 3. Radial basis function (RBF) neural network module:

[0046] The input dimension of the RBF module is the same as the number of input features, the number of BF layer nodes is 10, a Gaussian radial basis function is used, the number of nodes of the output layer is equal to the number of satisfaction categories 2, the width is determined according to the maximum distance between RBF centers and the number of nodes, the K-Means clustering method is used to determine the RBF center in the experiment, K-Means randomly selects k=2 points as the clustering center, assigns the clustering center to each sample through the Euclidean distance, the iteration number n_iter is set to 100, the gradient descent method is used to train the output layer weight, and the error between the predicted output and the real label is minimized.

[0047] 4. The classification tree module:

[0048] criterion=gini / entropy indicates that the splitting criterion is the Gini impurity / information gain; max_depth=10 indicates the maximum depth of the tree to prevent model overfitting; min_samples_split=5 indicates the minimum number of samples required for internal node splitting; min_samples_leaf=8 indicates the minimum number of samples required for leaf nodes; max_features=2 indicates the maximum number of features considered when finding the best split feature; random_state= controls randomness so that the results are repeatable.

[0049] 5. Control block:

[0050] The search space of the hyperparameter alpha is set, and the best parameter is found by traversing all possible parameters, and finally alpha=0.6 is selected to balance the MLP and RBF features in the experiment.

[0051] 6. Experimental evaluation index:

[0052] The accuracy rate is used to evaluate the satisfaction of online open course learners in the application, and the greater the accuracy rate, the more accurate the model satisfaction prediction result, and the accuracy rate is defined as follows:

[0053]

[0054] Among them, TP (true positives) represents true positives, which represents samples that are predicted to be positive and are actually positive; FP (false positives) represents false positives, which represents samples that are predicted to be positive but are actually negative; FN (false negatives) represents false negatives, which represents samples that are predicted to be negative but are actually positive; TN (true negatives) represents true negatives, which represents samples that are predicted to be negative and are actually negative.

[0055] 7. Method for predicting online open course learner satisfaction:

[0056] The method comprises a data acquisition and processing module, a multi-layer perceptron (MLP) module, a radial basis function (RBF) neural network module, a classification tree module and a control block, learner characteristics and satisfaction classification can be obtained online, and learner satisfaction prediction is realized through the acquired data.

[0057] 8. A method, device and electronic equipment for predicting learner satisfaction of online open courses, comprising: a data acquisition and processing module, configured to acquire learner characteristics and satisfaction classification, generate a learner satisfaction prediction dataset, perform data augmentation preprocessing to obtain an expanded satisfaction prediction dataset, and further divide the expanded satisfaction prediction dataset into a training set, a validation set and a test set.

[0058] a model construction and training module, configured to build a learner satisfaction prediction model of online open courses, train the model using the training set, and verify the model using the validation set, so as to obtain a trained learner satisfaction prediction model of online open courses;

[0059] a detection module, configured to input the test set into the trained learner satisfaction prediction model of online open courses to obtain predicted learner satisfaction.

[0060] The learner satisfaction prediction model of online open courses comprises the data acquisition and processing module, the multi-layer perceptron (MLP) module, the radial basis function (RBF) neural network module, the classification tree module and the control block.

[0061] 8. A method, device and electronic equipment for predicting learner satisfaction of online open courses, and a memory in communication connection with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the steps of the above-mentioned method for predicting learner satisfaction of online open courses.

[0062] A method, device and electronic equipment for predicting learner satisfaction of online open courses, which automatically records behavior data of learners through an LMS, an online interactive platform and a learning resource access log, and performs the steps of the above-mentioned method for predicting learner satisfaction of online open courses.

[0063] In an embodiment, a method, device and electronic equipment for predicting learner satisfaction of online open courses, the storage medium stores a computer program, and the computer program is executed by the at least one processor to perform the steps of the above-mentioned method, device and electronic equipment for predicting learner satisfaction of online open courses.

[0064] The storage medium includes Qt and OpenCV, the learner satisfaction prediction system is established through Qt; OpenCV is used to process data, realize human-computer information interaction, and implement the steps of the above-mentioned prediction online open course learner satisfaction method, device and electronic equipment.

Claims

1. A method, apparatus and electronic device for predicting MOOC learner satisfaction, the method comprising: The network framework includes a data collection and processing module, a multi-layer perceptron (MLP) module, a radial basis function (RBF) neural network module, a classification tree module, and a control block, and the method comprises the following steps: Step 1, the data acquisition and processing module first obtains the user's log data from the learning management system, including the learner's page browsing history H i , course progress P i , online test results R i , learning duration T i , the number of course interactions I i , then use the questionnaire to obtain the satisfaction of students online learning, Likert scale 5-point option to evaluate the satisfaction of respondents, 1 means very dissatisfied, 3 means neutral attitude, 5 means strong satisfaction, in the experiment, the label of 5 points is set to 1, indicating satisfaction, and the labels of 1-4 points are set to 0, indicating dissatisfaction, finally, the learning behavior data obtained is cleaned, standardized and normalized by using data preprocessing function; Step 2, the multi-layer perceptron module receives and analyzes the learning behavior data of the students, and performs deep learning through multiple hidden layers h to extract complex patterns in the learning behavior data and generate a satisfaction prediction based on the learning behavior, the model including an input layer, multiple hidden layers, and an output layer, each layer having corresponding weight and bias parameters, and being trained through a backpropagation algorithm; Step 3, a radial basis function neural network module configured to measure input data x using a Gaussian radial basis function i The distance d(x j ,c i ) between the preset center c j is used to measure the distance between the learning behavior data and the preset center, the model includes an input layer, a radial basis function layer and an output layer, the prediction performance of the model is optimized by adjusting the center and width parameters of the radial basis function, and the satisfaction prediction is generated based on this; Step 4, the classification tree module determines which data points are more suitable for the prediction results of the MLP model and which data points are more suitable for the prediction results of the RBF model, thereby improving the accuracy of the overall prediction; Step 5, Control block to integrate satisfaction prediction results from MLP and RBF models and and considering other possible data sources (e.g., course evaluations, feedback, etc.), output the final satisfaction prediction S i to provide timely and effective feedback to course providers and educational institutions. 2.The method, device and electronic equipment of claim 1, wherein, Online learning data of students is mined from a learning management system (LMS), and a data collection and processing module is used to implement data preprocessing functions for cleaning, standardizing, and normalizing learning behavior data obtained from the LMS to ensure accurate input for subsequent neural network models, the data preprocessing steps including outlier processing, missing value filling, and feature standardization to ensure data quality for model training. 3.The method, device and electronic equipment of claim 1, wherein, The MLP neural network model has an activation function: selecting a hyperbolic tangent function Tanh, a Sigmoid function, or a ReLU function; multiple hidden layers: each hidden layer includes multiple neurons for implementing multi-level abstraction and feature extraction of learning behavior data, and the output formula of the MLP model is as follows: where x ik is the kth input feature of the ith learner, w jk is the weight between the jth hidden neuron and the kth input feature, w j is the weight between the output neuron and the jth hidden neuron, b j and b h are the biases of the hidden and output layers, respectively, φ is the activation function, and f is the activation function of the final output.

4. The method, device and electronic equipment for predicting the satisfaction of online open course learners according to claim 1, wherein, The RBF neural network model uses a Gaussian radial basis function as a basic calculation unit for measuring input data x i The distance d(x j ,c i ) between the preset center c j ) is calculated according to the following formula: wherein σ j is a width parameter of the radial basis function, and the centers c j of the RBF neural network model are determined using a clustering algorithm, such as K-Means clustering, to optimize the performance and prediction accuracy of the radial basis function, The output formula of the RBF model is as follows: where β j is the weight of the jth radial basis function and φ is the radial basis function. 5.The method, device and electronic equipment of claim 1, wherein, The classification tree is used to obtain the prediction model to which the data points belong, and the output features and label data of the MLP and RBF models are configured, and the result is the features of each sample used for training the model. 6.The method, device and electronic equipment of claim 1, wherein, The control block integrates the prediction satisfaction S from the MLP and RBF neural network models i , and is divided into the ways of combining first and back propagation second and back propagation first and combining second, and the formula is as follows: Wherein, α is a parameter for balancing the output weights of the MLP and RBF models.

7. The method, device and electronic equipment for predicting the satisfaction of online MOOC learners according to claim 1, wherein, Data collection and processing module, first from LMS to obtain the student online learning behavior data, including but not limited to including the page browsing history H i , course progress P i , online test results R i , learning time T i And the number of course interactions I i , then use the questionnaire to obtain the satisfaction of students online learning, Likert scale 5 points option to evaluate the satisfaction of respondents, 1 indicates very dissatisfied, 3 indicates neutral attitude, 5 indicates strong satisfaction, finally using data preprocessing function to clean, standardize and normalize the learning behavior data obtained, so as to obtain the online learning behavior data of learners; The cleaned data is input into the MLP neural network model configured to receive and analyze the learning behavior data, identify and analyze the complex behavior patterns of the learners through the multi-level neural network structure, and generate the first-stage satisfaction prediction features The cleaned data is input into the RBF neural network model configured to receive and analyze the learning behavior data, measure the distance between the learning behavior data and the preset center through a Gaussian radial basis function, and generate the satisfaction prediction features in the second stage The classification tree module determines which data points are more suitable for the prediction results of the MLP model and which data points are more suitable for the prediction results of the RBF model; Integrating satisfaction prediction features from MLP and RBF neural network models using a controller and Consider other possible data sources (e.g., course evaluations, feedback, etc.), The final satisfaction prediction S is generated using the classification tree i . 8.The method, device and electronic equipment of claim 1, wherein, The memory, the processor, and the computer executable instructions stored on the memory and executable on the processor, wherein the processor executes the computer program to implement the online open course learner satisfaction prediction method of claims 1-6. 9.The method, device and electronic equipment of claim 1, wherein, The computer program is stored in the memory, and when the processor executes the computer program, the online open course learner satisfaction prediction method of claims 1-6 is implemented.

Images