Redefining the role of investors in South Korea’s startup ecosystem: A network analysis and explainable AI approach

Startups are vital drivers of innovation and economic growth, particularly in post–catch-up economies such as South Korea. This study investigates how heterogeneous investor types engage with startups through differentiated behaviors and network positions. Using the Innoforest database containing data on 338 investors and 3,570 startups, we construct an investor–startup network and apply a model-driven explainable AI-based machine learning approach to classify investor types based on capital, investment activity, and network centrality metrics. The analysis reveals which features most strongly influence investor classification and how these correspond to strategic roles such as early-stage brokerage, hub coordination, and commercialization. The results show that accelerators, venture capital firms, and financial institutions occupy distinct structural positions and engage startups at different stages with varying objectives. This study advances understanding of investor heterogeneity by linking structural positioning to behavioral roles and by illustrating how institutional logics in post–catch-up contexts shape investment engagement within startup ecosystems.