时间：2025年12月23日（星期二）下午14:00

地点：会计学院 301

参加对象：要求博士研究生参加，欢迎其他师生参加！

主题：Mapping the Hidden Supply Chain Network: A Machine Learning Approach

主讲人：陈世强博士 澳门大学

摘要：This study develops a new way to uncover global supply chain networks using unstructured financial news. Traditional data sources only report a small fraction of major customer–supplier links, leaving most commercial connections invisible. To address this gap, we collect more than 37 million news articles from 1979 to 2022 and build an annual firm-level co-occurrence network that reflects real interactions across industries and countries. We then apply graph neural networks to learn high-dimensional structural features for each firm, capturing both direct and indirect links in the business network. Using FactSet’s verified relationships as training labels, we combine graph embedding with supervised learning models to identify supplier and customer ties. The approach allows us to reconstruct supply chain networks beyond public disclosures, especially for small and medium-sized firms. Our findings show that news-based links provide rich information about economic relationships and offer a scalable method to map global supply chains. This framework also lays the foundation for future research on how shocks spread through production networks and how network structure shapes firm-level risk.