Abstract: 【Purpose/significance】Academic social network provides a platform for the interaction and sharing of research publica?
tions. The analysis of the characteristics of high-impact research publications in the platform will help expand the research dimension
of high-impact research publications and provide references for platform optimization and rational use of scholars.【Method/process】
This paper takes iSchool members as an example, selects 8449 high-impact research publications in ResearchGate, analyzes the char?
acteristics from the three perspectives of year, journal as well as author, and applies the time series clustering method to reveal the pat?
terns of influence changes.【Result/conclusion】The results show that most of the high-impact research publications are from recent
years, high level journals and showing high tendency of cooperation. Although there are some core authors, the author distributions are
still scattered. Some classical publications can also maintain their high impacts in academic social networks. The influence changing
mainly shows four patterns: linear growth, saturation tendency, recession tendency and explosive increasing. It reflects the overall
trend of high-impact research publications enhancing and keeping sustained influences through academic social network.【Innovation/
limitation】The innovation of this paper is to reveal the characteristics of high-impact research publications in multiple dimensions,
and use the time series clustering analysis method to summarize the change rules of indicators, so as to enrich the research on user be? havior of academic social network based on resource level.