Structured Dynamics for Multi-Behavior and Sequential Recommendation

Abstract

Sequential And Multi-Behavior Recommendation is advancing through efforts to align performance with evidence quality, resource limits, and transfer across settings. The review develops an evidence-centered account of combining relational supervision with dynamical models of stable preference, momentum, and abrupt change. Its analysis connects 2 focal papers with 11 independently retrieved publications screened through Crossref or the named publisher. The analysis is organized around behavior graphs, contrastive learning, temporal dynamics, interest decay, and offline evaluation. The analysis declines to treat metrics from unlike protocols as commensurate, the review compares problem definitions, methodological assumptions, and validation boundaries. Across the literature, a robust inference is that advances in sequential and multi-behavior recommendation become credible when representation, objective, and evaluation protocol are evaluated together and when uncertainty about distribution shift is reported explicitly. The synthesis ties method selection to use-case risk and reveals common limits on generalization, and proposes a research agenda centered on auditable baselines, controlled perturbations, and replicable records.