Key points are not available for this paper at this time.
Session-based recommendation systems focus on capturing users’ evolving intents from short interaction sequences, yet they persistently face three key challenges: the difficulty in dynamically discriminating between short-term and long-term interests, the inherent trade-off between sequential modeling and relational dependency learning, and the pervasive noise and sparsity in real-world session data. To tackle these issues, we propose MRGE (Multivariate Relationship Graph Embedding), a novel framework that synergizes enhanced recurrent modeling with graph-structured representations. Specifically, MRGE leverages a self-attention–enhanced RNN to concurrently model short-term intents and long-term preferences within sessions, while constructing a heterogeneous session graph that captures multi-relational item dependencies without compromising temporal fidelity. In addition, we introduce an auxiliary edge augmentation mechanism based on neighbor similarity to mitigate data sparsity and noise, thereby facilitating more robust information propagation. Extensive experiments on three public benchmarks— Delicious , Gowalla , and Foursquare —show that MRGE consistently surpasses state-of-the-art baselines and achieves significant improvements in top- \(K\) recommendation accuracy. Our implementation is available at: https://github.com/July-jz/MRGEcode .
Lou et al. (Mon,) studied this question.