Data-driven personalized marketing strategy optimization based on user behavior modeling and predictive analytics:...

Area of Science:

• Artificial Intelligence
• Machine Learning
• Data Science

Background:

• Personalized recommendation systems face challenges modeling complex user-product-query interactions.
• Existing methods struggle to effectively capture heterogeneous information networks.

Purpose of the Study:

• To propose a novel framework, DP-GCN (Deterministic Policy Graph Convolutional Network), for improved personalized recommendations.
• To integrate multi-level Graph Convolutional Networks (GCNs) with Deep Deterministic Policy Gradient (DDPG) reinforcement learning.

Main Methods:

• A graph-based embedding module captures multi-relational structures.
• A fusion layer integrates dynamic and static user and item features.
• A reinforcement learning layer adaptively updates recommendation policies based on user feedback.

Main Results:

• DP-GCN consistently outperformed state-of-the-art baselines on benchmark and real-world datasets.
• Significant improvements were observed in AUC, Precision@K, and NDCG@K metrics.
• The model demonstrated enhanced accuracy and adaptability in personalized recommendations.

Conclusions:

• Combining graph-based relational modeling with reinforcement learning is effective for personalized recommendation systems.
• DP-GCN offers a robust solution for complex user-product-query interactions.
• The framework improves both the accuracy and adaptability of modern marketing systems.