Abstract: 【Purpose/significance】The opening and sharing of scientific data can promote scholarly communication, where the aca?
demic virtual community provides a good basic platform for the opening and sharing of scientific data. This research will summarize
the communication mode of scientific data in the academic virtual community to provide a reference for the efficient exchange and
sharing of scientific data.【Method/process】We used qualitative analysis methods and social network analysis based on data exchange process to compare the features and communication mode for 8,521 pieces of dataset in GitHub.【Result/conclusion】This approach found that scientific data can be divided into four categories: raw data materials, analysis process data, model and theoretical data, and code test data. In the communication process, raw data materials are easy to obtain, easy to save, and easy to be discussed, but difficult to update. Analysis process data has the characteristics of being easy to save and easily update, but difficult to discuss. Model and theo? retical data has the features of being easy to save and be discussed, but difficult to update and be recognized. Code test data is most ap? preciated dataset by users. Moreover, in the academic virtual community, the mutual communication mode among users is relatively simple and straightforward. However, it has formed a communication pattern that surrounding scientific data. The circulation communi? cation mode prompts both scientific data users and academic virtual communities to improve the quality of scientific data production and dissemination.【Innovation/limitation】The proposed circulation flow of scientific data in academic virtual community will play an important role in and benefit scholarly communication。