DiffCold: A Diffusion-based Generative Model for Cold-Start Item Recommendation

DiffCold is a newly proposed diffusion-based generative model designed to tackle the cold-start item recommendation problem by unifying warm and cold item representations. It introduces a Retrieval-enhanced Aggregator and a Simulation-based Representation Alignment module to maintain distributional consistency and enhance the quality of generated embeddings. Experimental results demonstrate that DiffCold outperforms existing state-of-the-art methods, effectively addressing the seesaw dilemma that plagues traditional recommendation systems.