External and internal collaboration networks are crucial factors that encourage enterprises to engage in radical innovation activities. Exploring the complex mechanisms through which the characteristics of external and internal collaboration networks affect radical innovation performance is essential for innovative development by enterprises.
From the perspective of interaction between enterprises' external collaboration networks and internal inventor collaboration networks, this study examines the combination of key influencing factors and multiple improvement pathways for the radical innovation performance of Chinese artificial intelligence (AI) enterprises in heterogeneous collaborative contexts using machine learning methods, such as a K-means clustering algorithm and a classification and regression tree algorithm.
Based on the heterogeneity of collaboration network characteristics, Chinese AI enterprises can be divided into three types: externally oriented, internally extensive and internally cohesive. Hence, the characteristics of their external and internal collaboration networks and radical innovation performance differ significantly. The results reveal that the characteristics of both external and internal collaboration networks jointly influence enterprises' radical innovation performance, and the characteristics of internal collaboration networks play a more crucial role. Additionally, external and internal collaboration networks have complex nonlinear effects on enterprises' radical innovation performance through different combinations of characteristics.
This study reveals diverse pathways through which Chinese AI enterprises can enhance radical innovation performance in different collaborative contexts, offering insights into how to optimize the configuration of external and internal collaboration networks to achieve the strategic objectives of high-level innovation.
