Abstract:Sequential Recommendation (SR) captures users' dynamic preferences by modeling how users transit among items. However, SR models that utilize only single type of behavior interaction data encounter performance degradation when the sequences are short. To tackle this problem, we focus on Multi-Behavior Sequential Recommendation (MBSR) in this paper, which aims to leverage time-evolving heterogeneous behavioral dependencies for better exploring users' potential intents on the target behavior. Solving MBSR is challenging. On the one hand, users exhibit diverse multi-behavior patterns due to personal characteristics. On the other hand, there exists comprehensive co-influence between behavior correlations and item collaborations, the intensity of which is deeply affected by temporal factors. To tackle these challenges, we propose a Personalized Behavior-Aware Transformer framework (PBAT) for MBSR problem, which models personalized patterns and multifaceted sequential collaborations in a novel way to boost recommendation performance. First, PBAT develops a personalized behavior pattern generator in the representation layer, which extracts dynamic and discriminative behavior patterns for sequential learning. Second, PBAT reforms the self-attention layer with a behavior-aware collaboration extractor, which introduces a fused behavior-aware attention mechanism for incorporating both behavioral and temporal impacts into collaborative transitions. We conduct experiments on three benchmark datasets and the results demonstrate the effectiveness and interpretability of our framework. Our implementation code is released at https://github.com/TiliaceaeSU/PBAT.

What problem does this paper attempt to address?

The paper aims to address the challenges in Multi-Behavior Sequential Recommendation (MBSR). Specifically, the paper focuses on the following issues: 1. **Personalized Multi-Behavior Patterns**: Different users exhibit various behavior patterns on e-commerce platforms, such as clicking, liking, adding to cart, and purchasing. These behavior patterns vary from person to person, so using a unified approach to model all users' behaviors is not appropriate. Capturing personalized multi-behavior patterns is a challenge. 2. **Multi-Level Behavior-Aware Collaboration**: There are diverse collaborative relationships between the same type of behaviors. For example, in the behavior transition from "adding to cart" to "purchasing," adding a phone to the cart may prompt the user to purchase a Bluetooth headset later (complementary products), while adding headphones may reduce the probability of purchasing earplugs (competing products). This behavior sequence implies potential cooperative item relationships, which in turn affect the impact of behavior transitions. 3. **Impact of Time Intervals on Collaboration Strength**: The positional distance between behavior pairs directly affects their relevance. The smaller the interval, the stronger the collaboration. However, few studies have explored how time intervals directly affect behavior-aware collaboration. To address these issues, the authors propose a method called Personalized Behavior-Aware Transformer (PBAT). PBAT mainly consists of two modules: a personalized behavior pattern generator and a behavior-aware collaboration extractor. The former captures users' personalized behavior patterns through dynamic representation encoding and personalized pattern learning; the latter enhances recommendation effectiveness by extracting multi-level behavior-aware collaboration and considering the impact of time intervals. Experimental results show that this method outperforms existing methods on three benchmark datasets.

Personalized Behavior-Aware Transformer for Multi-Behavior Sequential Recommendation

Multi-Behavior Sequential Transformer Recommender

Multi-Grained Preference Enhanced Transformer for Multi-Behavior Sequential Recommendation

Behavior Pattern Mining-based Multi-Behavior Recommendation

Multi-Behavior Generative Recommendation

Recommender Transformers with Behavior Pathways

Time-Aware Preference Recommendation Based on Behavior Sequence

BPMCF: Behavior Preference Mapping Collaborative Filtering for Multi-behavior Recommendation.

Multi-behavior collaborative contrastive learning for sequential recommendation

A Survey on Multi-Behavior Sequential Recommendation

Multi-Behavior Hypergraph-Enhanced Transformer for Sequential Recommendation

A Multi-Behavior Recommendation Method Exploring the Preference Differences among Various Behaviors.

Modelling Temporal Dynamics and Repeated Behaviors for Recommendation

Multi-Behavior Sequential Recommendation with Temporal Graph Transformer

SSE-PT: Sequential Recommendation Via Personalized Transformer

Temporal-Aware Multi-behavior Contrastive Recommendation

Hierarchical Projection Enhanced Multi-behavior Recommendation.

Self-Supervised Enhancement Method for Multi-Behavior Session-Based Recommendation

Modeling multi-aspect preferences and intents for multi-behavioral sequential recommendation

BMLP: Behavior-aware MLP for Heterogeneous Sequential Recommendation

Two-stage Sequential Recommendation Via Bidirectional Attentive Behavior Embedding and Long/Short-term Integration