A Survey on Cross-Domain Sequential Recommendation

• DOI: 10.48550/arxiv.2401.04971
• Publication Date: 2024-01-01
• Authors: Shu Chen; Zitao Xu; Weike Pan; Qiang Yang; Ming Zhong
• Keywords: Survey research

Cross-domain sequential recommendation (CDSR) shifts the modeling of user preferences from flat to stereoscopic by integrating and learning interaction information multiple domains at different granularities (ranging inter-sequence intra-sequence single-domain cross-domain). In this survey, we first define CDSR problem using a four-dimensional tensor then analyze its multi-type input representations under multidirectional dimensionality reductions. Following that, provide systematic overview both macro micro views. From view, abstract multi-level fusion structures various models across discuss their bridges for fusion. focusing on existing models, basic technologies explain auxiliary technologies. Finally, exhibit available public datasets representative experimental results as well some insights into future directions research in CDSR.