High-Impact COVID-19 Research on Social Media: A Multi-Metric Study of Attention, Citation, and Topic Evolution

How to enhance the efficiency and quality of science communication, especially in the context of the continuous threat of global infectious diseases and the resurgence of the COVID-19 pandemic, has become a key issue in current publication evaluation and information governance. This paper conducts a systematic analysis of highly concerned COVID-19 research papers on social media by integrating methods such as bibliometrics, altmetrics, and text mining, revealing their dissemination characteristics, influence, and thematic evolution trends, and providing reference materials for the dissemination and evaluation of scientific information in the event of a possible major crisis in the future. The primary data source is the Dimensions COVID-19 literature database. The Altmetric Attention Score (AAS), an indicator of online impact, is used as the measure of social media attention. The research process begins by ranking papers in descending order based on their AAS scores and selecting the top 6,000 articles for analysis. Empirical methods such as descriptive statistics, correlation analysis, and Latent Dirichlet Allocation (LDA) topic modeling are then applied. The findings are as follows: (1) The top contributing countries of high-AAS papers are identified, along with the key research contents of the top ten papers and the major platforms on which these papers are shared. Differences in influence are also discussed in terms of author, institution, and country-level collaboration. (2) There is a positive correlation between citation counts, AAS, and journal H-index. Both the journal H-index and SJR (SCImago Journal Rank) are positively correlated with AAS, but AAS appears to have a stronger effect on the journal H-index than on the SJR. (3) Text mining results show that the 6,000 high-AAS papers cluster into four main research topics, with identifiable temporal evolution patterns. This study is based on a large sample of COVID-19 papers with high social media attention, integrating Altmetrics , traditional bibliometrics and text mining methods, and has achieved substantial expansion in research scale, analysis dimension and subject depth compared with existing research, providing a new paradigm for public health crisis communication research. The research results can provide a reference for expanding the theory and practice of public health bibliometrics.