Cross Domain Sequential Recommendation

Cross-domain sequential recommendation aims to improve recommendation accuracy by leveraging user interaction data across multiple domains, addressing the data sparsity common in single-domain approaches. Current research focuses on integrating large language models, employing graph neural networks and state space models to efficiently capture sequential patterns and mitigate negative transfer between domains, often using techniques like contrastive learning and disentangled representation learning. These advancements enhance recommendation systems' performance, particularly for cold-start users and in scenarios with limited data per domain, leading to improved user experience and business outcomes in various applications.