Modeling Behavioral Intensity and Transitions for Generative Recommendation

TL;DR

This paper introduces BITRec, a generative multi-behavior recommendation framework that models behavioral intensity and transition patterns explicitly, leading to significant performance improvements on large-scale datasets.

Contribution

BITRec presents a novel structured behavioral modeling approach with hierarchical aggregation and transition encoding, addressing limitations of existing generative methods.

Findings

Achieves 15-23% improvements across multiple metrics.

Peak gains of 22.79% MRR on Tmall.

Significant performance boosts on large-scale datasets.

Abstract

Multi-behavior recommendation aims to predict user conversions by modeling various interaction types that carry distinct intent signals. Recently, generative sequence modeling methods have emerged as an important paradigm for multi-behavior recommendation by achieving flexible sequence generation. However, existing generative methods typically treat behaviors as auxiliary token features and feed them into unified attention mechanisms. These models implicitly assume uniform activation of dependencies among historical behaviors, thereby failing to discern differences in intensity or capture transition patterns. To address these limitations, we propose BITRec, a novel generative multi-behavior recommendation framework that introduces structured behavioral modeling through selective dependency activation. BITRec incorporates (i) Hierarchical Behavior Aggregation (HBA), which explicitly models behavioral intensity differences through separated exploration and commitment pathways, and (ii) Transition Relation Encoding (TRE), which encodes transition structures through explicit learnable relation matrices. Experiments on four large-scale datasets (RetailRocket, Taobao, Tmall, Insurance Dataset) with millions of interactions achieve consistent improvements of 15-23% across multiple metrics, with peak gains of 22.79% MRR on Tmall and 17.83% HR@10, 17.55% NDCG@10 on Taobao.