Efficient and explainable sequential recommendation with language model

Zihao Li, Lixin Zou, Chao Ma, Chenliang Li

Abstract: Highlights•A unified framework is proposed by generating both recommendations and corresponding personalized explanations in a single process.•A rescaling adapter and a Fast Fourier Transformer (FFT) adapter are developed for Parameter-Efficient Fine-Tuning (PEFT), achieving superior performance with only a 5% update in parameters.•Comprehensive experiments on three public datasets demonstrate that PLEASER outperforms nine competitive baselines in sequential recommendation and three baselines in explanation generation.