Utilizing Graph Neural Networks for Identifying Similar Securities

Identifying similar securities is a critical task in financial analytics, influencing diversification, risk assessment, and portfolio management. This study presents a novel framework combining ChatGPT and Graph Neural Networks (GNNs) to enhance the analysis of structured and unstructured financial data. The integrated model leverages semantic embeddings and historical financial metrics for superior clustering accuracy, achieving significant improvements in normalized discounted cumulative gain and F1 scores. By capturing nuanced relationships, the proposed framework offers a robust solution for financial decision-making.