Abstract:   How open-source collaboration empowers the high-quality development of international innovation networks is a critical issue for R&D innovation in virtual communities in the era of digital intelligence. Taking the GitHub open-source community as an example, this study analyzed the project data from January 1, 2019 to June 30, 2023. Through machine mining and data crawling, a total of 9,090 data entries from 606 projects were extracted and processed from 59,120 raw data points. Based on the quantitative empirical analysis, the study explored the influencing factors and mechanisms of innovation performance in virtual communities. The results showed that from a social network perspective, network cluster density positively contributes to software innovation performance. From a perspective of complementary collaboration, geographic diversity, structural diversity and openness diversity positively moderate the relationship between cluster density and innovation performance. In contrast to the emphasis on geographical proximity in offline physical communities, geographic diversity plays the most significant moderating role in open-source communities. Structural diversity highlights multi-core project governance structures, while openness diversity advocates highly open software licenses as part of institutional design. This study will enrich the theoretical understanding of social networks and complementary collaboration, and offer new insights for improving digital governance of open-source community.

Key words: open-source collaboration, Github open-source community, innovation performance, social network, complementary collaboration