Artificial intelligence facial recognition and voice anomaly detection in the application of English MOOC teaching system

As a direct and effective biometric technology that follows human life habits, facial recognition has gradually become a mainstream, stable and reliable recognition method in the process of further development of science and technology. Facial recognition is a kind of biometric authentication based on recognition technology that is an original biological characteristic. After collecting the biometric functions, use a computer for digital image processing and template matching to complete the process of facial recognition. In the face of MOOC, establish a high-quality resource sharing mechanism, use this opportunity to explore innovative education models, and truly feel the strong momentum of traditional Chinese higher education, which has greatly improved the quality of education in Chinese universities. In addition, in order to vigorously promote the internationalization of education in China, education scholars need to conduct a lot of in-depth research on MOOC. In this article, a demand-based testing method is used to establish a skin color distribution model for color image preprocessing, and then develop and construct according to the comprehensive analysis of the university English class in the MOOC platform implementation mechanism (including the establishment of basic principles of operation mechanism), thereby become a guarantee and support for curriculum and education quality evaluation system. This article combines the characteristics of educational practice and MOOC education, takes college English courses as an example, studies its application mechanism, builds a MOOC platform, and continuously enhances students' interest in learning English, aiming to provide a practical reference for the reform of Chinese college English education.


Introduction
In order to determine the basis of information, the face is not only an important part of people's body, but also a reexamination of various kinds of information. At the same time, people's identity information can be identified based on the image of the face (Adjabi et al. 2020). In the field of computer vision, for visual information, the face is a great contribution to the provision of information. At the same time, in image recognition technology, face recognition can directly have a decisive effect on information, especially in the process of digital image processing, it has a very important influence.
Before face recognition, face detection is an indispensable and important step, and it is also the basis for the smooth progress of face recognition. The main task of this part is to collect images or video files and process them with algorithms to obtain facial data, which can be compared with the database, so that facial recognition can be obtained to determine the position of the face (Klare et al. 2012). After that, it can be used for data, facial image processing, and other operations after preparation. Facial recognition is a very convenient and fast way to recognize natural persons. There is no contact with each other, no effort is needed, and it will not affect people's normal living habits and behaviors. And due to the continuous innovation of Internet technology, the efficiency, function and accuracy of facial recognition application systems are getting better and better (Park et al. 2014).
In the vigorous development of the global open education movement, the combination of education and information technology is an important means to realize the sharing of high-quality resources, and it is also an inevitable requirement for realizing the globalization of education. As a new media that promotes social public opinion, ideology and cultural exchanges, the Internet has a huge social influence that cannot be ignored (John 2015). The development of education is also closely related to it. Therefore, the formulation of the education development plan must conform to the trend of social development and attach importance to the deep integration of education and the Internet. MOOC refers to large-scale open online courses. Different from pure video education in the past, learners can not only study and communicate online, but also obtain credits and certificates through online exams. This has brought unprecedented influence to the traditional education model of higher education. Since 2012, MOOC has developed rapidly in the world and is regarded as a new educational revolution (Sanchez-Gordon and Luján-Mora 2018). This is the process of improving a well-known MOOC platform, and it is also a resource sharing platform that is gradually improved in China under the traditional Chinese education model. At the same time, it puts forward new requirements for credit MOOC, for the learning certification carried out on this platform, and for the education implementation of Chinese universities. In terms of it, it is a huge problem (Gallagher and Savage 2016). The rapid development of MOOC has brought opportunities and challenges to traditional higher education, but traditional universities will neither die nor be replaced. On the contrary, traditional Chinese higher education can take advantage of this opportunity to establish a high-quality resource sharing mechanism and explore innovative education models. This not only greatly improves the quality of university education, but also can further help achieve the goal of the internationalization of China's higher education (Niu et al. 2016).

Related work
Based on the skin color model, the literature has studied and tested a facial recognition system combined with neural network algorithm recognition (Salama AbdELminaam et al. 2020). That is, process the collected source images, determine the target, and then perform facial recognition and resolution. In order to better achieve more accurate results and increase the success rate, the effect images delivered by the acquisition device will be processed. In terms of geometry and light, including digital image processing, considering many situations, select the appropriate processing mode and degree. In the literature, in order to obtain a robust sound detection method, two sound detection methods have been improved (Zhang et al. 2019). The performance of the weighted learning speech detection method based on powerful functions is better than that of MFREQ-MLRTVAD. The sound detection method based on noise and signal classification is better than MFCC-SVMVAD. The literature studies the application process of the MOOC platform based on the English courses of the university and conducts preliminary analysis from system support to curriculum construction (Tawafak et al. 2020). The author believes that the MOOC platform is an opportunity as well as a subject for the current situation. Based on the comparative analysis of the MOOC learning model and China's online college English education model, the literature proposes an online college English education system based on MOOC, with the purpose of providing a foundation for improving online college English and constructing the basic framework of China's education system (Chen and Zhang 2013). The literature proposes a personalized learning model based on MOOC. The model is designed, and the personalized learning mode is actually applied to the curriculum. This survey uses interviews, questionnaires, observations, and educational experiments to verify the effects of individual learning models in specific curriculum applications. MOOC can effectively supplement the past teaching methods, effectively supplement students' desire and initiative to learn, can effectively make up for the shortcomings of traditional English teaching methods, and promote students' individual learning. Therefore, MOOC is suitable for English learning in universities. The literature is based on the theory of personal learning and attention, and adopts the method of combining classroom performance recognition and head posture recognition technology to design and develop a class attention evaluation system based on classroom performance and head posture to arouse students' class attention and verify its effectiveness (Coleman et al. 2019).
The literature proposes a human motion tracking system based on facial recognition (Zhao et al. 2016;Zhang et al. 2022). In order to extract a more pure tracking target, before automatically drawing the tracking target, an attempt will be made to segment the selected area. However, there are certain problems with image segmentation. The focus of the literature is the application of facial recognition technology in the sponge-like information interaction design. The final result reflects that facial recognition technology can play an important role in highly responsive information interaction design. Combining responsive information interaction design with facial recognition technology can achieve a better experience and more efficient information transmission. The literature has studied the deep learning algorithm technology including CNN convolutional neural network, DNN deep learning technology, RNN cyclic neural network and other algorithms, analyzes their advantages and disadvantages, and analyzes automatic scoring, and compares the characteristics of artificial intelligence technology in education (Dixit and Silakari 2021;Sangaiah et al. 2023). Under the guidance of the in-depth learning theory, the in-depth learning engine TensorFolow is used to build a deep neural network model based on the existing English skills training system data of the university, extract the scoring function, and continuously track the model to adjust the neural parameters to construct the network model. The literature comprehensively analyzes the educational data generated by the system through detailed investigations of related theories and technologies such as data mining and machine learning (Romero and Ventura 2010;Sangaiah et al. 2022). After 2011, the education data generated by the education evaluation system was collected, the data was analyzed and processed, and individual and group portraits of students were produced. Construct a research environment that predicts the qualifications of students, make programs in Python language, and preprocess data extraction, cleaning, and integration.
3 Artificial intelligence facial recognition and speech anomaly detection algorithm model of English MOOC teaching system 3.1 Artificial intelligence facial recognition algorithm

Algorithm model
Computer vision optimization of images for emphasizing useful data of images. Image noise reduction and highlighting include Fourier transform, frequency domain and other methods. Regarding the brightness, the overall enhancement and transformation of the image will bring different damage to the image information. The histogram equalization operation is performed in the source image partition to effectively avoid excessive contrast changes when the gray level in the image is uniform. In this process, it is more effective to protect the detailed data of the image.
The formula related to the enhancement of the brightness component: In addition to brightness, you also need to increase the saturation of the image. This part corresponds to the corresponding change in chromaticity in image adaptability after preprocessing. This operation can be associated with processing the brightness component according to the color situation. The chroma increase correlation: Under different light conditions, the brightness of the image may affect the predetermined variation range of the facial area. This article uses the adaptive light compensation method. The basic formula is as follows: In order to express the skin color for positioning, the first thing that needs to be established is the feature model of skin color. There are models of various skin tones that are applicable at this point. This includes Gaussian models and parts of these models that are improved by the application. Other ellipse models are models based on statistical histograms. In different color representation spaces, skin color features have different clustering functions, so the process of skin color modeling depends on the color space. Now, in the development of various platforms, there are widely used color spaces such as RGB images, YCRCB color space, HSV (HSI), and the lesser-known YIQ and CIEXYZ.
In the implementation process, the face detection using YCbCR color space is realized. This color space is also called YUV, which is a basic face detection color space. CB is the blue component, and CR is the red component. The dynamic collection of video streams belongs to the basic general form. The color space and HSI color space in the performance of the application have the same performance, but the scope of the application is different. Contrary to the widely used RGB representation, information is not lost during the conversion process (see Fig. 1). The biggest advantage of the YCBCR color space application used here is that the brightness value is a single parameter. This is useful when you need to separate luminance data.
The relevant formula is as follows: where the parameters are The skin color model is established, the color data structure in the image is mathematically operated, and the result data of the operation is the further binarization of the image, and the color range is obtained through such a process. The analysis effect of the skin color field is shown in Fig. 2: The speed of specifying the position of the face based on the skin color model is very fast, and the calculation level of this computer is very low, so the distinction of skin color is a good way to improve efficiency. If the light and shadow conditions and the conditions of the image acquisition device are ideal, the image position of this segmentation algorithm can be more effective than the algorithm based on the Haar eigenvalue. The skin tone similar method is only suitable for testing and stabilizing operations. The effect is very good in images where the position and area of a single face exceeds 10%, but the image detects more faces, smaller faces, and more skin tones. But the positioning effect is reduced.
PCA feature extraction is also called principal component analysis. The main principle of this method is to integrate the data scattered in the components into comprehensive and representative data in the case of high-dimensional data information such as image data that needs to be processed. Application methods of statistical analysis. In a mathematical sense, this is a simple and basic operation, which has the application value of data compression and dimensionality reduction. The matrix calculation to be applied is basically the calculation of the covariance matrix.
While obtaining a specific face sample, the partial space of the main elements is established through calculation, and the conversion component is executed. The representation vector selected in this process can also be expressed as an inherent face. Related formulas include the following PCA specific vector calculation formulas: BP neural network is defined as a multilayer feedforward network structure, and the training method of neural network is a backward error propagation algorithm. BP neural network is one of the most widely used neural network structures. This kind of neural network does not need to set up the mapping function in the first step, and can more conveniently perform machine learning and save operating parameters. As a typical artificial neural network, its network topology has several main components, including an input layer and an output layer, and a hidden layer containing one layer and multiple hidden node layers. This paper applies the Sigmoid function which contains arithmetic expressions. The formula of the sigmoid function is as follows.

Identification process
The algorithm flow in this paper is mainly image source acquisition, image preprocessing (including image emphasis, illumination correction, etc.), face positioning (color space modeling, region closure operation), and recognition algorithm. The simple flowchart is shown in Fig. 3: The basic process of positioning and detection is shown in Fig. 4: 3.2 Speech anomaly detection algorithm

Algorithm model
From a physiological point of view, from the lungs to the throat, then from the vocal cords to the vocal tract, sound waves are emitted from the mouth to form speech. Figure 5 shows the digital model generated by the sound signal. As shown in the figure, the speech generation system consists of the following three parts. (1) The excitation function produced by the glottis. (2) The modulation function produced by the sound channel. (3) The radiation function produced by the lips. These three functions are cascaded into the transfer function of the speech generation system, and the formula is as follows: The main idea of the sound detection based on the statistical model is to assume that the noise signal and the sound signal follow a specific statistical model distribution, and to calculate the parameters of the model corresponding to each frame signal (shown in Fig. 5). Finally, the detection is performed by the likelihood ratio. This method was originally proposed by Sohn. The discrete Fourier transform of a multi-noise audio signal is as follows: If two hypotheses are given: it means that the sound does not exist, it means that the sound exists. In this case, sound detection can be seen as the following binary hypothesis detection problem, that is: Assuming that both N(t) and S(t) follow the zero-average Gaussian distribution, the probability density function of the frequency component Y(t) under the assumptions of H0 and H1 is as follows: Artificial intelligence facial recognition and voice anomaly detection in the application… 6859 The likelihood ratio of LTCH frequency components is as follows: Here, n l t ð Þ and c l t ð Þ are the previous signal-to-noise ratio and the next signal-to-noise ratio, respectively, defined as follows: In the direct decision estimation method, the prior SNR is estimated using formula (23): Assuming that all frequency components are statistically independent, the geometric mean of the likelihood ratio of frame t is defined as follows: Among them, K is the number of frequency bands in the current frame.
The decision rule for speech detection is calculated based on the geometric mean of the likelihood ratios of the frequency components of each frame of the signal, so the decision rule is as shown in Eq. (25): In order to improve the detection of weak sound signals, a sound detection method based on multiple observed likelihood ratios is proposed in the literature. In this method, when determining the first frame, multiple frames before and after the frame must be used at the same time, and the number of data frames involved needs to be determined using sliding windows and statistical modeling. The likelihood ratio is executed for each frame of the data in the window, and the final decision rule is composed of the sum of the likelihood ratios. The establishment of the decision rule is expressed in formula (26) In recent years, researchers have proposed verification methods based on statistical hypotheses, long-term spectral difference methods, amplitude probability distribution methods, low-dispersion spectroscopy methods, etc., and many sound detection methods with excellent detection performance have been proposed. What these methods have in common is the use of background noise estimation or noise suppression as part of sound detection. Therefore, these methods usually have many control parameters, which limit the application environment of sound detection. As the difference in long-term spectra, VAD depends on the choice of seven control parameters of the method. Therefore, in order to avoid selecting control parameters, some researchers have focused on sound detection methods based on signal classification. This is constructed using VAD as a binary classification problem. The system block diagram is shown as in Fig. 6.

Detection simulation
The sound detection accuracy, false alarm rate, and detection error rate of the Factory, Babble, and Subway methods with different SNRs are shown in Tables 1, 2, and 3. Table 1 shows the accuracy of the voice detection method. Table 2 shows the false alarm rate of voice detection methods under different background noises. Table 3 shows the detection error rate of the voice detection method under different background noises.
From Tables 1, 2, and 3, it can be seen that the accuracy value of NoLTSPVAD is very close to the accuracy value of Hmfreq-MOLRTVAD. This phenomenon shows that just by weighting the geometric mean of Hmfreq-MOLRT to improve these scores does not have much impact on the accuracy of VAD detection. By adding the geometric mean based on LSP to the recognizable frame, it can be seen that LtSPVAD has higher detection performance than NOLT SPVAD. When the SNR is 0 dB and 5 dB of background noise, the LTESPVAD method is better than the other four reference VAD methods, and the corresponding geometric average weight based on LSP is very large. Obviously, the use of LSP can effectively improve the detection of weak sound frames. The false alarm rate and false detection rate of the proposed VAD are lower than the reference method.
4 Practical application of artificial intelligence facial recognition and voice abnormality detection in English MOOC teaching system

Design of English MOOC teaching system
Operating mechanism refers to the general term of various factors that affect the actions and relationships of test subjects. The development and construction of college English MOOC courses need to establish a flexible and efficient application mechanism in order to effectively complete the construction tasks. In the initial stage, policy support and financial investment from the school are needed. In the intermediate stage, team members must split their work, be responsible for the development of online courses, and establish an effective incentive mechanism and feedback mechanism. In the latter stage, to establish a scientific evaluation and supervision mechanism, all links must support and cooperate with each other. Among them, the incentive mechanism includes spiritual incentives and material incentives, and specific remuneration standards must be formulated. The feedback mechanism is operated by technical personnel. A special feedback window is set up on the course learning page. It is the responsibility of the corresponding personnel to collect and process special feedback (see Fig. 7).

Basic principles
The university's current education system is combined with the MOOC education model, and it is necessary to study the establishment of the MOOC operating mechanism for college English courses. Only through this method can the established operating mechanism be scientific and reasonable and help to play a positive role. However, the following principles also need to be followed: The classroom-centered traditional education model, teachers arranged according to the time of the teaching content, and students in classrooms that accept limited textbook knowledge are limited to limited space, and it is easy to overlook the dominant position of students in learning. In the development of large-scale online open courses, the educational model obviously cannot meet the learning needs of students. The reform of school education requires not only the reform of educational content, teaching methods and educational resources, but also the adjustment of the school's talent training plan and teaching plan. Of course, as a new form of MOOC, the construction of the curriculum system and the adjustment of educational procedures can provide effective help. Schools can independently design and develop their own MOOC courses, introduce excellent courses from other schools, enrich the educational content and improve the quality of education.
Another high-quality educational resource is a common problem for universities, which is detrimental to school management and professional training. Therefore, universities must accelerate the establishment of high-quality educational mechanisms, promote exchanges in resource sharing, inter-university exchanges and communication, and seek common development.

Conducive to improving the teaching ability of teachers
The teacher is the subject of the educational process, and the student is the subject of the learning process. Due to the emergence of MOOC, the direction of the teacher's role has changed. In the previous education model, teachers conveyed knowledge to students through ''face-to-face'' education in the class. However, due to the limitation of class time, the large number of students, and the difference in learning ability, teachers cannot consider individual students and cannot guarantee the quality of education. Thanks to the launch of MOOC, students can study individually outside the classroom, and students and teachers can enjoy the same educational resources. Therefore, teachers must improve their own qualities and abilities, deepen their understanding of knowledge, in order to help students in the class do good work that is exciting and helpful. The development of MOOC requires teachers to pay attention to cultivating students' autonomous learning ability, improve students' mastery and understanding of knowledge points, and help students improve their practical application ability. This is another request of the teacher. Educational reform must not only use modern information technology to reform the teaching method, but also explore the combination of online learning and reversal teaching. Teachers must continuously improve their teaching and academic research capabilities, and actively participate in the entire process of MooC education.

Conducive to meeting the learning needs of students
MOOC courses follow the student-centered principle, providing high-quality educational resources and simple operations. Students can choose courses freely according to the goals of the period, combining their own characteristics and the requirements of making a study plan. In order to learn, you may encounter problems that teachers and classmates cannot solve in time. In addition, you can also provide multiple resources to use the platform to solve your own problems. Through self-disciplined learning, students can control their own learning progress. You can see things they don't understand many times, and that is no longer limited by the time of the classroom and teachers. Through online communication and classroom discussions, students can deepen their knowledge and promote the development of personality. This innovative learning method of MOOC meets the personality and learning requirements of students, and has won many students, but in the learning process of MOOC, students must continue to learn, face the Artificial intelligence facial recognition and voice anomaly detection in the application… 6863 deficiencies, and be full of enthusiasm and participation. Make the most of online resources, complete tasks assigned by teachers, participate in class discussions and share ideas. According to the needs of future career development, students not only want to learn professional knowledge, but also want to improve their abilities. Use knowledge to solve problems and improve overall quality.

Teaching objectives and requirements
College English courses are compulsory courses for all schools. Teachers are required to follow the rules of language learning, adopt effective teaching methods, use the latest information technology, create a good learning environment, make full use of the advantages of online learning, and improve students' practical language application ability. The guidance of college English courses must meet the requirements of the society for talents. Modern college students must not only understand theoretical knowledge, but also combine knowledge and practice to create value. This is the purpose of learning. Therefore, in order to help students better integrate into society, it is necessary to emphasize practical courses in practical education. Not only schools, but teachers also need to help students change their backward ideas, encourage students to implement various learning activities, give play to their actual communication skills, and make English education more in line with the needs of student development. In the future, cultivating comprehensive talents useful to society is a demonstration of improving the quality and effectiveness of education.
The guidance of college English courses is studentcentered, focusing on cultivating students' independent learning ability, improving students' enthusiasm and ability to participate in learning, changing the past teaching methods, guiding students to participate in online learning, and enabling students to enter personalized learning surroundings. Online learning is convenient for students to choose their own learning and progress, with good grades, and the ability to accept students to complete research tasks quickly. After choosing to learn other related knowledge, the openness of resources allows students to enjoy the rich world of high-quality educational resources and a wide range of knowledge fields: Students with poor school performance and poor acceptance ability can watch the video many times until they understand, but they also have enough time to digest in order to avoid slow progress in the entire course.
In the process of college English learning, teachers must pay attention to guiding basic knowledge such as vocabulary and grammar. Regarding oral English, students need to communicate daily, and they need to use standard pronunciation and clearly express their opinions and opinions on specific issues. In the sense of listening, they can understand each other's meaning and respond through daily communication. Students can understand about 120 words per minute in slow English. When taking the listening test, students can use the skills they have learned to help solve problems. Regarding reading ability, general English reading ability can be read at a reading speed of about 65 words per minute, and the central idea of explaining them can be mastered. When writing, students can ensure the correctness of vocabulary and grammar and the fluency of the article. According to photos and text prompts, a short essay of 120 words can be written within 30 min.

Teaching methods and means
(1) MOOC online education. Students can visit the online learning website through the Internet, choose the course video to watch after logging in, answer the questions in the video to deepen their understanding, and complete the learning tasks to get results. In addition, you can also communicate with teachers and classmates online in real time to expand your knowledge.
(2) See the teacher in the class. After the students' online learning is over, the teacher will sort out the classroom discussions according to the learning content, help students answer questions, encourage students to exchange ideas, merge learning content, and allow students to participate in independent learning and class discussions to improve students' learning status and improve students' learning ability.
(3) Internship guidance. Teachers assign tasks and apply what students have learned to practice. The study of language courses requires a lot of oral communication. In fact, it can improve students' oral expression skills and cultivate the ability to participate in society.

Assessment method
Course evaluation is mainly based on online learning, faceto-face learning, social practice and final exams. The results of online learning are mainly obtained by watching videos and completing online exam questions and topics.
The results of face-to-face classes are mainly evaluated by the teacher based on the student's performance and class attendance. The so-called social practice is to comply with the practical requirements assigned by teachers. Students independently participate in practical activities after class, and after submitting a practice report, the teacher makes an evaluation. The result of the final exam is a unified suggestion put forward by the teacher, which constitutes the exam, and the study and exam method of the students this semester. The overall scores of the students in the above four aspects are different. According to the actual situation of the major, you can choose the proportion of teachers in each major, and the evaluation in the college English class. This article uses 40% online learning ? 15% classroom learning, and 15% ? 30% social practice. For the final test score, A \ 65 is not satisfactory, 65 or above is passed, 75 is average, 85 or above is good, and 95 or above is excellent. Regarding exam qualifications, it is stipulated that students who missed 3 times in a semester and failed to complete the homework 3 times are directly regarded as students who failed the exam and can take the exam again in the next semester.

Course content and schedule
For example, the new version of the Comprehensive College English Course (Volume 1) is the main textbook of college English. The guidance content of each chapter is independently designed, including background, vocabulary, sentence analysis, pre-class assignments, etc. For the courses corresponding to the composition of the article, all the class hours can be summarized in Table 4. According to the construction standards of XX University MOOC published by the university, the production of course video needs to be combined with the materials and recorded courseware provided by the teacher for recording and editing (Table 5). The specific requirements are as follows: 1. At the very beginning, the courseware needs the school's logo, course information, teacher information, etc.
2. The final content of the credits can be added according to the needs of the course. 3. Before the class, teachers need to introduce themselves and introduce the courses. 4. Teachers must pronounce clearly, with consistent speed and intonation. 5. After the role of the video is completed, the camera must run smoothly. 6. The inserted information must be consistent with the content of the lecture, and the content that is not related to the course must not be displayed. 7. The length of the video can be determined according to the content and speed of the lecture. 8. The recording environment is quiet and no noise is required (Table 6).
Because the course video needs to be published on the WisdomTree education platform, please note that the code stream of the video must be transcoded by a predetermined decoder before it is online.

Teaching quality evaluation system
The study of the MOOC education quality evaluation system makes a reasonable evaluation of all aspects of curriculum construction, finds out shortcomings and revises it, improves the construction level of online open courses, accelerates the pace of university education reform, and enables the sustainable development of  Chinese education. In addition to the success of the construction of the MOOC course, whether it is unique, the corresponding scale, excellent equipment, and sufficient equipment investment, it is also necessary to investigate whether the school's rules and regulations are sound and whether the teachers' educational ability is excellent. Whether the student accepts awareness or other questions. According to research, effective education basically depends on teachers' abilities, such as teachers' knowledge accumulation, educational skills, educational organization, and ability to communicate with students, etc., to achieve the expected educational effects, but student participation, concentration of learning, and existing Knowledge base. Therefore, the establishment of an education quality evaluation system needs to combine qualitative evaluation and quantitative evaluation to ensure that the evaluation indicators are consistent with the school's direction and development strategy, to better achieve the purpose of evaluation, promote sustainable development, and improve education and service levels. Due to the particularity of the MOOC education model, students learn through online self-study guidance and offline self-study. On the one hand, a school's education quality evaluation system is established based on the existing evaluation system. On the other hand, based on the particularity of MOOC education, it is very important for teachers to evaluate students' selflearning ability and take corresponding measures to solve problems. The establishment of an evaluation system (shown in Fig. 8) must be comprehensive and scientific in order to ensure the standardization of control measures. Evaluation content is generally obtained through questionnaire surveys, university feedback, student interviews, classroom observations, homework tests, completion reports and other methods to establish evaluation indicators. According to the flow chart of the evaluation system, firstly, teachers and students are surveyed by questionnaire survey and interview methods. According to the survey, the evaluation content is grasped, evaluation indicators are determined, and the implementation of the evaluation is organized. In the evaluation system, the evaluation of data, analysis of evaluation results, and the need to convey ideas to the school management department, and feedback to teachers and students.

Conclusion
MOOC is an opportunity for higher education in China, but also a challenge. In terms of opportunities, first of all, it can promote educational equity and increase the chance of receiving higher education. MOOC is Internet-based online courses, most of which are free, and most of them cooperate with famous universities to ensure high-quality education. Learners can adjust the time, choose courses, and take exams according to their own situation. The emergence of MOOC provides more learners with opportunities to receive high-quality education, which helps to improve the quality of education for ordinary people. Second, we must promote the sharing of educational resources and accelerate the internationalization of higher education. Because the MOOC platform is open to the world, wellknown universities in China can open excellent courses on the MOOC platform, so that students all over the world can learn online. On the other hand, Chinese students can also learn excellent courses from foreign universities, realizing the sharing of global high-quality educational resources. Finally, it helps to improve the quality of education and effectively combines online and offline for comprehensive learning. Regarding challenges, traditional Chinese higher education has also been affected, bringing greater challenges to some local universities and widening the gap between elite schools and ordinary universities, which will raise higher and higher teaching methods and teacher quality.