Identifying Chinese Leading Venture Capital Firms Based on Graph Convolutional Neural Networks

It is a meaningful challenge to identify leading venture capital firms (VCs) in the analysis of the Chinese investment market. Identifying leading VCs is equal to determine influential nodes in the field of complex network analysis. Many studies have applied centrality measures to determine influence nodes. However, only a few studies have explored more efficient and flexible ways to accomplish this task. In this work, we propose a new approach which using graph convolutional neural networks to identify influential nodes in the network, so as to determine leading VCs. We build an undirected graph based on co-investment of VCs, then learn a VCs Graph Convolutional Neural Network (vcGCNN) for nodes classification. Our vcGCNN is labeled with '1' and '0' for `is leading VCs' and 'is not leading VCs'. The experiment results on VCs dataset demonstrate that vcGCNN outperforms multiple centrality measures and some typical spectral-based GNN methods for leading venture capital firms identification.