Graph-Augmented Transformer Framework for Context-Aware Recommender Systems on Amazon E-Commerce Data

Recommender systems play a pivotal role in enhancing user experience and driving engagement across e-commerce platforms. However, traditional approaches such as collaborative filtering and matrix factorisation often struggle with sparse user-item interactions and fail to capture contextual nuances. To address these limitations, this paper proposes a novel hybrid framework that integrates Graph Neural Networks (GNNs) and Transformer-based architectures for context-aware recommendation. The framework first constructs a dynamic bipartite graph from user interactions and product metadata, enabling the GNN to learn relational embeddings that reflect user preferences and item similarities. These embeddings are then fused with a Transformer module that applies multi-head self-attention to model temporal and semantic patterns within user behaviour sequences. We propose a hybrid Graph-Augmented Transformer framework for context-aware recommendation, evaluated on the Amazon Reviews 2023 dataset (571M reviews, 54.5M users, 48.2M items). The model integrates graph-based relational embeddings with transformer-based sequential attention. Compared to LightGCN, SASRec, and BERT4Rec baselines, our approach achieves significant improvements in NDCG, recall, and robustness under cold-start conditions. The framework is scalable and suitable for real-time deployment, offering a practical blueprint for next-generation recommender systems. By combining graph-based relational learning with attention-driven context modelling, this research contributes a powerful and flexible solution for next-generation recommender systems.