Multi-Behavior Generative Recommendation

Zihan Liu, Yupeng Hou, Julian J. McAuley

Abstract: The task of multi-behavioral sequential recommendation (MBSR) has grown in importance in personalized recommender systems, aiming to incorporate behavior types of interactions for better recommendations. Existing approaches focus on the next-item prediction objective, neglecting the value of integrating the target behavior type into the learning objective. In this paper, we propose MBGen, a novel Multi-Behavioral sequential Generative recommendation framework. We model the MBSR task into a consecutive two-step process: (1) given item sequences, MBGen first predicts the next behavior type to frame the user intention, (2) given item sequences and a target behavior type, MBGen then predicts the next items. To model such a two-step process, we tokenize both behaviors and items into tokens and construct one single token sequence with both behaviors and items placed interleaved. Furthermore, we design a unified generative recommendation paradigm that learns to autoregressive generate next behavior and item tokens, naturally enabling a multi-task capability. Additionally, we exploit the heterogeneous nature of token sequences in the generative recommendation and propose a position-routed sparse architecture to efficiently scale up models under the generative recommendation paradigm. Extensive experiments on real-world public datasets demonstrate that MBGen significantly outperforms existing MBSR models across multiple tasks.