Abstract: Highlights•We model each view internally using methods tailored to its data characteristic.•We propose DCDL with diffusion learning to enhance recommendation diversity.•Extensive experiments on three public datasets validate DCDL’s effectiveness.
Loading