Political teaching application in high vocational care courses based on machine learning systems

Higher vocational colleges need to use video system technology to transform teaching methods to transform ideological and political teaching into more vivid, more attractive and infectious forms, thereby effectively assisting students to actively learn. This article studies a video system based on machine learning and has developed a type of high vocational nursing course political teaching system. The system uses the principle of machine learning to build a learning early warning monitoring algorithm, and then, the performance and accuracy of the model are evaluated by using the algorithm. This article finds that the module can extract the hidden characteristics of traffic and is not affected by the knowledge. The design of the high vocational nursing course political teaching system is equipped with multiple modules such as video on-demand, management, interaction, and recording. Compared with the group comparison, it is very efficient to find that the high vocational nursing course political teaching system in this article is very efficient. This article studies machine learning technology and applies it to the construction process of high vocational nursing curriculum political teaching video systems.


Introduction
Ideological theory is the key course in the training of talents in colleges and universities. The practical educational effect of such courses is closely related to China's strategy of strengthening the country, and it is also the key way to provide high-quality talents (Xu et al. 2021). Its teaching process and method affect the actual teaching effect and the training process of talents. Therefore, in the new era environment, how to do a good job of ideological and political teaching in higher vocational colleges is a major challenge in the field of education reform (He et al. 2021). In the Internet age, the use of multimedia technology and information technology can effectively change the teaching methods and make them more suitable for the current preferences of students (Xiao et al. 2019). Optimizing the teaching process by combining relevant technologies can improve the attraction of the classroom to students, thus cultivating information technology professionals, which is also one of the important objectives of education reform. The main channel for students to receive ideological and political education is higher vocational education, which needs to be based on mobile Internet technology to meet the actual needs of students from the background of the new era (Li and Fu 2020). Keep pace with the times, change according to the situation, and adopt new media and new technologies to make the classroom environment more vivid. In recent years, the video teaching method represented by hybrid learning method has appeared and gradually developed (Smith et al. 2012). Compared with single online teaching or traditional teaching methods, video system teaching has unique advantages. It can combine traditional teaching with video courses, so as to carry out collaborative learning, structured learning and unstructured learning (Soucisse et al. 2017). Based on this, this paper studies the video system based on machine learning and applies it to the political teaching system of higher vocational nursing courses. At the level of students, the system can improve their learning level, stimulate their & Zixia Zhou 11904@jsmc.edu.cn enthusiasm for learning, and cultivate their professional ability and information literacy. At the teacher level, the system can help ideological and political teachers in higher vocational colleges not only to have a deeper understanding and thinking of the subject knowledge and professional skills, it is necessary to continuously strengthen the application of information technology, so as to more reasonable teaching planning to improve the teaching effect, improve students' learning enthusiasm. In view of the management personnel in the teaching process, it is necessary to summarize advanced teaching experience and skills, so as to better manage and allocate teaching resources and make relevant decisions, on this basis to complete teaching innovation and transformation, and improve the overall teaching effect and quality of colleges and universities. At the same time, it is necessary to exchange information with other colleges and universities, learn from each other's strengths, achieve common progress, and promote the deeper application and practice of video teaching system.

Relevant work
The literature emphasizes the management optimization of educational administration system, which requires such systems to actively communicate with foreign schools to promote information interaction while following the operation and design standards (Yi and Fang 2019). The system needs to be regularly updated and maintained, including innovative talent training mode and active information service mode. Then, the construction technology and performance requirements of the educational administration management system are analyzed (Clarke and Kouri 2009). The literature specially studied the software engineering course, and through the construction of good characteristics, combined with the mixed learning environment data of the University online learning platform and offline classroom, the early warning learning task was scored (Zhang and Huang 2021). As for knowledge points and problem types, the ''knowledge point feature network ? other feature network'' and ''problem type feature network ? other feature network'' based on the neural network model are designed to deal with the two types of features, respectively, which can solve the early warning problems of high granularity problems and score the early warning problems (Sharma et al. 2017;Yakubu et al. 2020). Aiming at the learning withdrawal warning task based on students' course selection and learning data in the teaching network of a certain university, the literature summarizes the course selection scenario as the regression classification problem class and uses the good integration effect to learn the classification problem of numerical type. The lightgbm method model clustering strategy solves the problem of low accuracy in the classification warning problem to a certain extent (Yang et al. 2021;Sangaiah et al. 2023). A semi-supervised feature selection algorithm (srr-lsa) based on improved ReliefF and ACO algorithms is proposed in the literature. The improved ReliefF algorithm improves the limitation that the original ReliefF algorithm only evaluates the importance of features, can quickly reduce the size of features, and can provide good prior knowledge for the improved ACO algorithm. During feature extraction, candidate features are randomly combined by ReliefF algorithm for weight calculation (Sheikhpour et al. 2017). The algorithm uses Spearman correlation coefficient to perform correlation analysis and preliminarily eliminate redundant features. In this paper, a combination of feature engineering input model and fixed length time series packet is proposed for deep learning to identify encrypted traffic. In addition to some characteristics indicating the flow and information in the packet header, the feature vector of the packet is also added to the first 60 bytes of the data layer of the application program (Zhao et al. 2008). The packet vector group is a sequence of 5 packets, which are, respectively, used as a token. Then, CNN is used to learn spatial features and LSTM is used to learn time series features. The model weights the output probabilities of unencrypted and encrypted modules to obtain the final probability estimation (Sangaiah et al. 2022). Experimental results show that the feature selection algorithm can effectively reduce the complexity of feature subset, achieve better classification effect, and make the system itself have better generalization and classification ability (Wang et al. 2016;Sangaiah et al. 2021). A target classification model is derived in the literature. This process is realized by using bagging-c4.5 training samples. This sample will generate similar C4.5 base classifiers in the actual model training process, so it will increase the training time and reduce the accuracy. It needs to use classification algorithms such as genetic algorithms to integrate the base classifiers to reduce the training time (Bar-Hillel and Weinshall 2008;Sangaiah et al. 2020). When predicting the graduation status, the system needs to use C4.5 decision algorithm and naive Bayesian classification to train the model, improve the classification accuracy of the model, and finally, design a reasonable strategy to obtain the final prediction model (Saritas and Yasar 2019).
3 Research on key technology of machine learning

Traditional machine learning theory
The return of Logic Slim is a simple linear model, which is easy to achieve. It is a classification model represented by conditional probability distribution. It is often used in the field of classification and recommendation. As shown in the formula (1).
Among them, x is the input feature, and w is the feature weight. In other words, the output of the number of likeminded models is represented by the input linear model.
The target function of XGBOOST is shown in formulas (2) (3).
The goal is to make the weight of the network similar to the weight of the real model. First, initialize the connection weight and the value of each unit and multiply by the random number, and then, enter the training data for network parameter optimization, and finally, get the actual results of each layer. x Calculate the training error based on the difference between the actual output value and the expectation value, as shown in formula (5) and formula (6).
Reverse propagation learning right values and training target functions are shown in formulas (7) and formulas (8).
Among them, g is the learning rate. Equation is the expected error at the end of the q round of training. E is the sum of the expected errors of all rounds.

Learning behavior monitoring and early warning algorithm
The biggest difficulty of using Bayesian to seek the probability after seeking the probability is that the conditional probability of a class is the combination probability of all attributes, but it is difficult to directly obtain its value in a limited data sample. The simple Bayesian algorithm needs to be used for classification. This process is independent. For the training data set, the characteristics are independent of the characteristics, and the input and output data probability research is performed based on this point. Then, the calculation of the postponent probability of the Bayesian theorem through the sample classification of the model, and the highest value category results as such samples as such samples.
Among them, C is a category of category set c about C. There are only two types of graduates and non-graduates in the study of graduation state. n is a characteristic number of parts of the data set, and xi is the characteristic attribute value of the I of the sample x. C's probability. For each category, P (x) is the same, so according to the Bayesian criteria above, we get the formula classification of simple Bayesian: It can be seen from the above formula that training a simple Bayesian classifier is to use the data concentrated data to calculate the initial probability of each class and the condition of each attribute. The formula for calculating the prior probability of a class is as follows: Calculation of conditional probability is related to data types. If the characteristic attribute data are a discrete value, the conditional probability is the i -i value of the characteristic attribute of the characteristic attribute of the class tag of the training data set. The total number of samples | Dc, xi | Class tags as the ratio of samples in C | Dc |, the formula is as follows: If the characteristic attribute is continuous data, according to the probability density function, assuming the condition probability obeys the normal distribution, the calculation formula is as follows: Political teaching application in high vocational care courses based on machine learning systems 7659 When using the simple Bayesian attribute weighted algorithm to obtain the student's prediction model, first of all, according to the effects of the characteristics on the classification results, to obtain the appropriate weight, and then, use the weighted coefficient to fix the simple Bayesian formula. Use the simple Bayesian formula to obtain the weight, defined as the formula (13): Information gain is usually used to select the attributes of the segmentation characteristics in the decision tree algorithm. In information acquisition, the more information brought by the characteristic properties to the classification system, the greater the importance of the characteristic attribute to the entire classification system. The so-called amount of information is the information entropy, so this chapter uses the information amount brought by the feature attribute to the system as the basis for judging the importance of characteristic attributes and determines the weight according to this.
GainðD; AÞ ¼ HðDÞ À HðD j AÞ ð 14Þ The calculation formula of the experience entropy and condition entropy in the formula is as follows: HðD | CK | It is the number of samples that belong to CK data sets. K is the number of categories in category C. M is the number of attributes of attribute A. | Dj | is the number of samples in the subset Dj. The number of samples of the subset Dj of CK.
The higher the information value obtained by the characteristic attribute A, the more impact on the final classification results of the students' graduation warning research, and the greater the importance in the classification. Calculate the information value obtained by each attribute according to the formula and then, get the weight value of each feature attribute. The formula is as follows: When the classification results of the two classification models are merged, the test sample is calculated based on the accuracy rate of the two models in the accuracy of the data set and the post-test probability of the two models and the corresponding post-verification probability of the two models.
C is a fixed category value. AWNB is the final accuracy of the simple Bayesian classification model when classifying the data set. AC4.5 is the final verification of the classification tree model, that is, the accuracy of the data set in its classification Essence P (c | x) WNB is obtained by all simple Bayesian classifiers. The classification result of the sample classification is the proportion of class C classifiers in all classifiers. Calculate as follows: Among them, N is the number of simple Bayes classifiers, and R (T) is a classification output. C4.5 Decision Tree Classification Model P (c | x) C4.5 Calculation is similar to simple Bayesian.
Calculate imbalance. Express the only sample number of the student data set as ns, the number of samples of most categories in the future as nl, and the formula for calculating imbalance is as follows: Synthesized as: For each sample in the unique sample, the Euclidean distance formula is used to calculate the k nearest neighbors of the sample, D Is the number of samples of the highest category in the K neighborhoods of the sample. The ratio R is calculated as follows: Step 3 obtains the ri of each sample in a few classes and standardize the ri. Calculated as follows: Calculate the number of samples to be synthesized in each minority sample in the calculation data: Select a minority sample xzi from each neighbor of each of the small samples xi to each of the small samples to be synthesized and generate a new sample based on the following formulas, where the k is random number, k [ [0,1].
Pulphic synthesis until the number of new samples meets gi.

Overall model parameter selection
This article selects different N as a candidate feature collection for experiments. The feature selection algorithm further adopts the above SRR-LSA algorithm. In summary, the application layer data head contains important information such as protocols and encryption algorithms. Such information usually has a greater impact on the identification and classification of encrypted flows. Therefore, this article selects the top 300 words of the application layer. This part is based on 10 bytes, defined as {h1, h2, …, h30}. After many experiments, the selected features are {h1, h2, h3, h4, h5, h6}. In addition, the accuracy of the test concentration is shown in Fig. 1. In the experimental results of the first 100 bytes, N that is 60 is the local maximum value. The selection of the selection algorithm consistent with the results is added to the input feature vector of the deep learning model.
Since the model adopts a two-stage identification method, the final output depends on the fusion probability estimation of the unencrypted part and the encrypted part. The output weight of the encryption module is: q. The output ratio of the unencrypted module is 1-q. According to the results of the adjustment algorithm, the parameters q 0.77. Table 1 shows the differences q model recognition performance of the test data under the value.
When the value is 0.77, the accuracy rate reaches the maximum value. Because the non-encrypted module is mainly aimed at stream-level characteristics, the encryption module analyzes and recognizes the content of the stream-level content by extracting the load characteristics of the data packet. Although encrypted traffic data cannot directly represent application information, the encryption algorithm and protocol contained in the Baotou description are still distinguished. At the same time, deep neural networks can extract the hidden part of traffic and are not affected by the knowledge of priority, so the weight of the module is higher.

System infrastructure design
According to the system function analysis, this article designed a functional module of the high vocational care course based on machine learning video systems. The system consists of video demand modules, management modules, interactive modules and record modules (Fig. 2). Among them, realize the functions of video on-demand, video content guidance, video index; the management module performs the loading, modification and deletion of video resources, the addition, modification and deletion of the column, and the management of the forum. Interactive modules perform maintenance functions such as online interactive and learning forums; record modules realize functions such as displaying students' basic information, learning process records, publishing topics and posts, upload experiences, and uploading experiences in the form of personal learning files.

Learning early warning system module design
The overall business process of the system is divided into two paths, as shown in Fig. 3, which represents the overall process from early warning data pre-processing to training tuning, and finally, to the front-end display. User browsing systems, watching video and other learning behaviors enter the database through user data management modules. The system calls the content of the database by the early warning learning module, and returns the processing results to the user through the warning window, which mainly includes the warning result of the return and exit of the warning score result. The administrator calls the content of the database management curriculum and early warning learning through the curriculum management module and the learning early warning module. The purpose of the final model is to learn early warning, so it is not enough to predict the results. This work needs to know the current warning level of students, divided into red warning (R), yellow warning (Y), and green warning (G), depending on the warning, depending on the warning. The level is divided into three levels, and the division rules are shown in Table 2. Teachers can target students at different levels of early warning and learning according to the final warning learning level provided by early models.

System simulation results analysis
After completing the testing of the political teaching system of the higher vocational nursing course, the test results were analyzed and studied, and the output sample category was compared with the actual sample category to evaluate the performance of the model. This article experimented with student data collected at the end of the first semester of the first semester of the first grade, the end of the second semester and the end of the second semester of the second grade, and the target data was political test data. Compare the test results of different time points with the actual results, and calculate the accuracy, recall and accuracy rates, respectively. Compared with the results of the simple Bayesian classification model and the Bagging-C4.5 classification model, the result is shown in Table 3: Oracle11 database server resource usage is shown in Fig. 4: The usage of web server resources is shown in Fig. 5. Judging from the above test results, the database server and web server are always in a stable state and basically meet the relatively stable requirements of the system during online operation. At the same time, the higher the number of online people, the higher the number of online personnel. It can be seen through the results of the pressure test that the number of online users is about 4000, and the system is relatively stable when the concurrency is issued at about 20%.

Learning improvement effect test
As shown in Table 4, of which cooperative learning and thinking the scores of political information literacy have been significantly improved.
Comparing the scores of the total self-learning ability and four dimensions before and after traditional ideological and political teaching, after statistical inspection, the difference is not statistically significant, P [ 0.05, (as shown in Table 5).  Fig. 2 The module structure of the system Fig. 3 The overall business flowchart of the system  Political teaching application in high vocational care courses based on machine learning systems 7663 5 The application research of the multimedia video system in political teaching

The application effect of high vocational political teaching in video systems
This article is to verify whether the political teaching system of high vocational nursing courses based on the video system can achieve the expected effect, and the verification effect of the system needs to be analyzed. The ultimate purpose of the analysis is to check the performance of the system and whether it can meet the needs of users, and debug and change the system according to the results to provide users with more benefits. As shown in Table 6, it was found in the survey feedback that 47.5% of the students thought that the use of video system was helpful to carry out ideological and political education, and 36.4% of the students strongly agreed. In the survey on the advantages of using video system to publicize ideological and political education, 66.7% of the students thought that using clear audio and video to display the theme and creating a teaching environment with context could enable students to feel the impact of new knowledge, thus stimulating students' enthusiasm for learning and ensuring continuous learning. 80% of the students think that the system can provide different types of video resources, keep up with the times, share knowledge, broaden their horizons, and be easily accepted by students. 86.7% of the students think that in the same online interaction, students can express their opinions frankly and without worry, which makes it easier for teachers to understand their feelings. Among them, 83.3% of the students think that the system can solve the   problems of insufficient ideological and political teachers and uneven teaching levels, and 65% of the students think that the network environment is conducive to self-study.

Promotion strategies of political teaching in higher vocational nursing courses
First of all, it is necessary to optimize the teaching content, adjust the teaching tasks and teaching topics. New teaching methods should be selected, case discussions should be flexibly used, film and television works should be watched, and courseware made by students should be displayed and exchanged, so as to stimulate the interest and enthusiasm of teachers and students to actively participate in teaching, and strive to form comprehensive evaluation forms including online and offline evaluation, summative evaluation and formative evaluation. By changing, optimizing and changing the traditional classroom teaching form, make it more suitable for the teaching requirements. Then, it is the system construction of the ''network classroom'' of Ideological and political courses in higher vocational colleges. First, the network teaching platform of interactive learning and management is selected. After the online teaching platform is determined, online courses can be built by themselves, or the existing course resources of the platform can be used to build online classroom teaching of Ideological and political courses in the school through the mixed teaching mode of ''MOOC resources ? smart classroom campus ? offline,'' and organize real-time testing, selection, question and answer, discussion, examination and other teaching activities. And manage, detect and evaluate the whole learning process. The online classroom construction of political courses should be based on the principles of simplicity, practicality, convenience and stability, and effectively ensure the development of teaching quality.
Then, to effectively change the ''practice classroom'' of Ideological courses, we need to develop and improve the existing teaching system. The practice platform relies on multimedia technology, database, network communication, virtual simulation and other technologies, carefully designs and deeply develops cultural resources, and organizes students to implement network expansion guidance.
Through the construction of such a practical three-dimensional classroom with standardized management, rich content and diverse forms, we can promote its integration with the traditional classroom and the network classroom, so as to realize the unity of knowledge and action.
Teachers should upload micro-class videos, teaching cases, teaching courses, test questions and other teaching materials related to the classroom to the online cloud classroom. Students will conduct self-study in the online classroom according to the needs of teachers. After completing their homework, they can interact with other students or teachers. Through the online teaching system, teachers can timely obtain the students' learning progress and problems in the learning process so as to adjust the teaching tasks and difficulties. At this stage, teachers uploading resources and students' autonomous learning are important components of blended learning.
With interactive learning as the core, the links usually involved include summary, case analysis, student discussion and achievement display, so as to help students further consolidate the knowledge they have learned. The teaching activities at this stage should be carried out in the classroom equipped with information-based teaching equipment.
In the aspect of classroom presentation, the cooperative group spontaneously recommends some students to report their learning achievements in the form of oral report or PPT before the whole class and answer questions to show their learning achievements. Subsequently, the teachers will organize some students to communicate and discuss the problems in the display stage. Case analysis is a discussion topic put forward by teachers for the case, so as to cultivate students' ability to apply knowledge and stimulate their innovative consciousness. The corresponding summary is the summary of each chapter and the teacher's comments. This point is a key link in the mixed learning mode and has a specific guiding role. Due to the great difference between students' understanding level and knowledge level, there are errors in the submitted homework. Teachers should give specific guidance and correct the errors in time. Finally, teachers should establish logical connections between knowledge points, analyze and summarize key and difficult points, show students the context of knowledge in the form of graphics, help students clarify their ideas and improve learning efficiency.

Conclusion
The political teaching system of higher vocational nursing course based on ''video machine learning system'' is different from the traditional teaching concept of ''teaching'' and ''learning.'' The system combines online teaching, Political teaching application in high vocational care courses based on machine learning systems 7665 video teaching and classroom teaching. While helping students learn knowledge, it emphasizes the empowerment of students, so that students can use practical skills while learning new knowledge and cultivate political literacy. Through the active application and practice of teaching, the political teaching of nursing courses in higher vocational colleges has effectively promoted the reform of classroom teaching, improved the teaching quality and efficiency, it has greatly improved the enthusiasm of higher vocational students to study ideological and political theories and has made certain achievements, thus has a significant impact on students, teachers and even the society. We hope to continue to deepen this applied research and practical exploration, truly create an ideological and political course that is loved by college students and benefits all their lives, and make it a course with a higher professional level. Thus, it provides China with more comprehensive talents and can better practice the development of the cause with socialist characteristics.
Funding The research was supported by: (1)  Data availability Data will be made available on request.

Declarations
Conflict of interest The authors declare that they have no conflict of interests.
Ethical approval This article does not contain any studies with human participants performed by any of the authors.