Detecting High-Potential SMEs with Heterogeneous Graph Neural Networks
Researchers have developed SME-HGT, a Heterogeneous Graph Transformer framework designed to identify Small and Medium Enterprises (SMEs) with high potential for advancing in funding rounds. The model utilizes public data to predict which Small Business Innovation Research (SBIR) Phase I awardees are likely to secure Phase II funding. By constructing a graph of companies, research topics, and government agencies, SME-HGT achieved an AUPRC of 0.621, outperforming baseline models and demonstrating significant lift over random selection. AI
IMPACT This research demonstrates a novel application of graph neural networks for identifying promising SMEs, potentially aiding investors and policymakers in resource allocation.