ACADEMIC RESEARCH • 學術研究 2022 UMAGAZINE 26 • 澳大新語 65 were reopened to resume face-to-face teaching. In addition to the content of the microblogs, we also collected the profiles of their creators, including their gender. To compare the microblogs from the three time periods, we used Bidirectional Encoder Representations from Transformers (BERT) for sentiment analysis. Specifically, we used this machine learning technique to automatically classify the microblogs into three sentiment categories: positive, negative, and neutral. Each microblog was tagged with a numerical sentiment value of +1, 0, or - 1 to indicate positive, neutral, or negative polarity. For content analysis, we used the Latent Dirichlet Allocation (LDA) algorithm to identify the topics to which a piece of unstructured text data belongs. We found that the number of microblogs about online education increased dramatically during the Mid-pandemic Period, peaking at 430,566 microblogs in March 2020, when schools and other educational institutions in mainland China completely suspended face-to-face classes and switched to online learning. This number is over ten times higher than that from the first month of our study (33,415 in July 2019) and over 40 times higher than that from the last month of our study (9,366 in November 2020). This indicates that the emergencies that started in January 2020 caused a lot of concern about online education among the population, especially among women. Indeed, the dramatic increase in the number of microblogs about online education during the Mid-pandemic Period came mainly from female users. In the first and third periods, most male and female users seemed to be equally indifferent to online education. Before the pandemic, 79.03 per cent of microblogs in the study were classified as neutral towards online education, and only 4.67 per cent of them were negative. However, in the Mid-pandemic Period, only 33.04 per cent of the posts were classified as neutral, and 51.63 per cent of them were negative. Moreover, the proportion of female users who held negative views was much higher in the third period (9.27 per cent) than that in the first period (3.26 per cent). This implies that the pandemic has had a lasting negative impact on women’s attitudes towards online education. During the Mid-pandemic Period when online education was the only alternative to face-to-face teaching, Weibo users generally held negative attitudes towards online education. This is consistent with a study of Chinese users’ satisfaction with selected online education platforms during the pandemic, but contrasts with previous studies 在三個時間階段有關線上教育的微博帖文數量 The total number of microblogs about online education during the three periods 在三個時間階段有關線上教育的微博帖文數量（按用戶性別劃分） The total number of microblogs about online education during the three periods (by gender)
RkJQdWJsaXNoZXIy MTQ1NDU2Ng==