ChronoID: Infusing Explicit Temporal Signals into Semantic IDs for Generative Recommendation

ChronoID introduces a novel framework for integrating explicit temporal signals into semantic IDs for generative recommendation systems, addressing the limitations of time-agnostic models. The framework systematically characterizes the design space for temporal signals and includes a new benchmark for evaluating time-aware recommendation performance. This approach is significant for practitioners as it aligns model architectures with real-world temporal dynamics, potentially enhancing recommendation accuracy and relevance.