Empowering recommender systems using automatically generated Knowledge Graphs and Reinforcement Learning

2023-07-11

arXiv.org

AI Summary

To address low user engagement and insufficient decision interpretability in personalized financial service recommendations, this paper proposes an end-to-end interpretable recommendation framework integrating knowledge graphs (KGs) and reinforcement learning (RL). Methodologically: (1) it automatically constructs a multi-source, heterogeneous financial KG to unify structured and unstructured data representations; (2) it introduces a Path-Directed Reasoning (PDR) mechanism leveraging deep RL (DQN/Policy Gradient) for traceable, explainable path discovery over the KG; and (3) it adopts a dual-track interpretability paradigm—XGBoost-SHAP-ELI5—to jointly ensure high predictive accuracy and feature-level attribution. Empirical evaluation on real-world financial datasets demonstrates significant improvements in click-through rate (+23.6%) and average session duration (+18.4%). Moreover, the framework delivers regulatory-compliant, auditable recommendation justifications, establishing a novel paradigm for intelligent investment advisory systems and customer relationship management.

Abstract

Personalized recommender systems play a crucial role in direct marketing, particularly in financial services, where delivering relevant content can enhance customer engagement and promote informed decision-making. This study explores interpretable knowledge graph (KG)-based recommender systems by proposing two distinct approaches for personalized article recommendations within a multinational financial services firm. The first approach leverages Reinforcement Learning (RL) to traverse a KG constructed from both structured (tabular) and unstructured (textual) data, enabling interpretability through Path Directed Reasoning (PDR). The second approach employs the XGBoost algorithm, with post-hoc explainability techniques such as SHAP and ELI5 to enhance transparency. By integrating machine learning with automatically generated KGs, our methods not only improve recommendation accuracy but also provide interpretable insights, facilitating more informed decision-making in customer relationship management.

Problem

• Enhancing personalized article recommendations
• Integrating Knowledge Graphs with Reinforcement Learning
• Improving interpretability and decision-making in CRM

Innovation

• Knowledge Graphs generation
• Reinforcement Learning traversal
• XGBoost with explainability