JEPA4Rec: Learning Effective Language Representations for Sequential Recommendation via Joint Embedding Predictive Architecture
This work addresses data sparsity and generalization issues in sequential recommendation systems, particularly benefiting scenarios like cross-domain, cross-platform, and low-resource settings, though it appears incremental as it builds on existing language representation and JEPA methods.
The paper tackled the problem of data sparsity and limited understanding of common-sense user preferences in sequential recommendation by proposing JEPA4Rec, a framework that combines Joint Embedding Predictive Architecture with language modeling of item descriptions, resulting in improved recommendation performance and reduced reliance on large-scale pre-training data, with experiments showing it consistently outperforms state-of-the-art methods on six real-world datasets.
Language representation learning has emerged as a promising approach for sequential recommendation, thanks to its ability to learn generalizable representations. However, despite its advantages, this approach still struggles with data sparsity and a limited understanding of common-sense user preferences. To address these limitations, we propose $\textbf{JEPA4Rec}$, a framework that combines $\textbf{J}$oint $\textbf{E}$mbedding $\textbf{P}$redictive $\textbf{A}$rchitecture with language modeling of item textual descriptions. JEPA4Rec captures semantically rich and transferable representations, improving recommendation performance and reducing reliance on large-scale pre-training data. Specifically, JEPA4Rec represents items as text sentences by flattening descriptive information such as $\textit{title, category}$, and other attributes. To encode these sentences, we employ a bidirectional Transformer encoder with modified embedding layers tailored for capturing item information in recommendation datasets. We apply masking to text sentences and use them to predict the representations of the unmasked sentences, helping the model learn generalizable item embeddings. To further improve recommendation performance and language understanding, we employ a two-stage training strategy incorporating self-supervised learning losses. Experiments on six real-world datasets demonstrate that JEPA4Rec consistently outperforms state-of-the-art methods, particularly in cross-domain, cross-platform, and low-resource scenarios.