Article Preview
TopIntroduction
With the full penetration of the Internet and the rapid expansion of mobile Internet, almost all traditional industries will be affected, and “Internet+” is changing one traditional industry after another. In recent years, influenced by the global COVID-19 pandemic, the online education craze has reached a new height. In the new form, the “Internet + traditional education” that is Massive Open Online Courses (MOOC) has become the main path for people to acquire knowledge and improve themselves. The opportunity is that the Internet can deepen the model of traditional education reform, bringing new markets and new demands (Zhu et al., 2020). The challenge is that with the rapid and continuous development of MOOC, it faces urgent problems; firstly, how to set more effective content to attract more students and increase the flow of MOOC sites; secondly, online learning is different from the traditional face-to-face learning mode of teachers and students; thirdly, how to detect the learning effect of students; and fourthly, how to provide more effective guidance during the learning process of students. Finally, how to establish more effective communication with students online and understand their learning process has become an urgent problem for each platform (Dai et al., 2020). From the perspective of users, researchers can efficiently and accurately conduct emotional analysis of online education platform comments, help potential users quickly find the information they need, screen false and exaggerated information, and make wise choices in the case of complex and asymmetric information. Users can select products according to their own needs and characteristics, which greatly reduces the time and energy spent by users in selecting products. From the perspective of product operators, combined with the potential needs and emotional tendencies of users for the product mined from the in-depth learning model, find the pain points of the product and the direction of iterative development, better improve the service experience of the product, enhance the user experience, and enhance user stickiness (Martin et al., 2020).
With the development of artificial intelligence, the collected user information can be processed through big data analysis, data mining, and other technologies to discover the association and knowledge connotation, which provides us with many technical improvements and even significant scientific discoveries. However, many deep-learning and machine-learning methods require high data requirements, and in order to ensure user privacy, MOOC platforms often encrypt the data and are unable to extract data such as user learning videos and operations. Therefore, seeking less private data to complete user analysis is a prerequisite to ensure the security of the platform (Paudel, 2021). Currently, most open data in each platform are course evaluation messages, so how to use these data, analyze course preferences, improve course settings, and understand user habits is important for the development of the platform and for students. For the information mining of messages, it can be abstracted as the intelligent text processing problem in natural language processing, and the flowchart for such a problem is shown in Figure 1.
Figure 1. The flowchart for the intelligent text analysis
For students, their most intuitive feelings will be displayed in the form of text comments after completing the course, and these text comments contain their thoughts and emotions about the course, so how to effectively complete this kind of information extraction is the key to improving the quality of the MOOC platform (Mujahid et al., 2021). The essence of sentiment processing of text comments is natural language processing technology, and with the development of deep learning, the use of neural networks to complete the sentiment classification and feature extraction of text has become the most important research approach (Strubell et al., 2019). Therefore, for the problem of sentiment analysis in online education platforms, this paper uses deep neural networks to perform sentiment discrimination so as to complete the rapid screening of positive and negative emotions, to maximize the utility of data, and to improve the strategy formulation process. The specific contributions of this paper are as follows: