Abstract:    This paper explored the hierarchical mechanisms through which innovation network embeddedness, network structure, and network content influence corporate technological innovation. By integrating the patent data, alliance network data, and financial data from Chinese biopharmaceutical firms, we constructed the panel data and employed the social network analysis methods and fixed-effects negative binomial regression models to examine the impact of alliance network embeddedness, network structure, and network diversity on corporate innovation performance. The results revealed that: (1) Local embeddedness, overall embeddedness, structural holes, and diversity of alliance types all positively influence corporate innovation performance; (2) Structural holes exhibit varying moderating effects across different levels of embeddedness; and (3) Alliance type diversity negatively moderates the relationship. By introducing the PageRank algorithm, the study confirmed differentiated mechanisms at local and whole network levels during the innovation process. The embeddedness mechanism highlights the importance of strong connections and network cohesion while the structural hole effect underscores the strategic value of network break and weak ties in the network. Moreover, the diversity effect suggests that network complexity hinders innovation benefits from the network. Therefore, managers, when formulating network innovation strategies, should distinguish between ego and whole network strategies, and incorporate information about relationships between alliance partners and types of alliances into the strategic decision-making process.

Key words: alliance network, network embeddedness, structural hole, innovation performance, PageRank algorithm