Abstract:With the development of text mining and analysis techniques, social media can gain more comprehensive and objective insights into news text. In this paper, more than 110,000 blog posts published on Weibo platforms by nine official mainstream media from January 1, 2021 to April 13, 2022 were obtained and analyzed by LDA-based topic modeling. Taking the presentation and construction of science communication and science education topics in news as the entry point, this paper relies on the theory of online agenda setting to identify and sort out the science topics in news , and obtain the distribution of science topics in each media and the contribution of each media under each science topic. It is found that all blog posts can be grouped into five advanced themes and twenty initial themes. The Covid-19 theme and the science and technology education theme among the five themes have strong scientific attributes, and the eight initial themes under them revolve around scientific topics such as sudden public health and safety events, prevention of telecommunication fraud, and aerospace. The percentage of blog posts under these eight science communication and science education-related themes ranges from 20% to 50% of the nine major media, and the proportion of specific themes also has its own media characteristics. The contribution distribution by each media under different science topics is both common and unique, reflecting the collaboration and reasonable allocation of science topics among media. In general, the agenda setting of science issues on social media by the nine major media is clearly contemporary, responsive, and adaptive. The media respond to the public''s demand for scientific information in a timely manner and provide efficient and accurate science communication and science education by combining the public''s current information acquisition habits.
