Publication Type

PhD Dissertation

Version

publishedVersion

Publication Date

4-2025

Abstract

The financial industry operates within a highly dynamic and interconnected ecosystem, presenting unique challenges for predictive modeling and decision-making. Accurately forecasting financial performance, assessing credit risk, detecting fraud, and ensuring compliance require methodologies that can capture complex temporal, relational, and contextual dependencies within financial data. This thesis investigates the use of Temporal Relational Graph Convolutional Networks (TRGCNs) combined with financial knowledge graphs (FKGs) to address these challenges and enable advanced analytics in the financial domain. We introduce FintechKG, a financial knowledge graph constructed through a threedimensional information extraction process, incorporating entities, temporal dimensions, and domain-specific financial relationships. A TRGCN-based framework is proposed to model both temporal and relational dependencies within FintechKG, enhanced by FinBERT embeddings and social media data integration for richer feature representations. The framework is evaluated using financial performance prediction as a case study, where a logistic regression model is applied to classify revenue trends, demonstrating the effectiveness of combining relational and textual embeddings. Additionally, this thesis presents a Temporal Credit Knowledge Graph (TCKG) framework for corporate credit risk assessment. By employing temporal embeddings, relational reasoning, and generative AI techniques, we establish a comprehensive methodology for credit risk prediction. A comparative analysis of five predictive models—including logistic regression, Relational Graph Convolutional Networks (RGCNs), generative AI-based Gemini models, and Graph Recurrent Neural Networks (GRNNs)—reveals the superiority of the Gemini-based approach in terms of accuracy, robustness, and interpretability. By integrating temporal learning, relational reasoning, and generative AI, this research advances the state of the art in financial analytics. It provides a systematic framework for constructing financial knowledge graphs and demonstrates their transformative potential in addressing critical challenges in the financial sector. This work not only highlights the efficacy of TRGCNs for financial applications but also paves the way for future research into integrating knowledge graphs and generative AI for enhanced decision-making in dynamic financial environments.

Keywords

Temporal Knowledge Graphs, Graph Neural Networks, Credit Risk Assessment, Relational Graph Convolutional Networks (RGCN), Financial Knowledge Graphs, Generative AI in Finance, Anomaly Detection, Financial Performance Prediction, Multimodal Data Integration, Explainable AI (XAI)

Degree Awarded

Doctor of Engineering

Discipline

Graphics and Human Computer Interfaces | OS and Networks

Supervisor(s)

FANG, Yuan; DAI, Bingtian

First Page

1

Last Page

148

Publisher

Singapore Management University

City or Country

Singapore

Copyright Owner and License

Author

Download

Share

COinS