Multiplex Behavioral Relation Learning for Recommendation via Memory Augmented Transformer Network

Abstract

Capturing users' precise preferences is of great importance in various recommender systems (e.g., e-commerce platforms and online advertising sites), which is the basis of how to present personalized interesting product lists to individual users. In spite of significant progress has been made to consider relations between users and items, most of existing recommendation techniques solely focus on singular type of user-item interactions. However, user-item interactive behavior is often exhibited with multi-type (e.g., page view, add-to-favorite and purchase) and inter-dependent in nature. The overlook of multiplex behavior relations can hardly recognize the multi-modal contextual signals across different types of interactions, which limit the feasibility of current recommendation methods. To tackle the above challenge, this work proposes a Memory-Augmented Transformer Networks (MATN), to enable the recommendation with multiplex behavioral relational information, and joint modeling of type-specific behavioral context and type-wise behavior inter-dependencies, in a fully automatic manner. In our MATN framework, we first develop a transformer-based multi-behavior relation encoder, to make the learned interaction representations be reflective of the cross-type behavior relations. Furthermore, a memory attention network is proposed to supercharge MATN capturing the contextual signals of different types of behavior into the category-specific latent embedding space. Finally, a cross-behavior aggregation component is introduced to promote the comprehensive collaboration across type-aware interaction behavior representations, and discriminate their inherent contributions in assisting recommendations. Extensive experiments on two benchmark datasets and a real-world e-commence user behavior

Graph Meta Network for Multi-Behavior Recommendation

“…• NGCF 𝑀 [29]: we enhance the NGCF model to inject the multibehavioral graph relations under a graph neural network. • MATN [35]: it differentiates the relations between user and item with the integration of the attention network and memory units.…”

Collaborative Reflection-Augmented Autoencoder Network for Recommender Systems

“…Many efforts have focused on proposing recommendation models with different types of neural networks [2,4,11,43], such as neural network-augmented recommendation via stacking several feed-forward layers [12,14] and neural auto-regressive learning frameworks [6,55]. More recently, several graph neural networks (e.g., graph convolutional network) have been applied to recommendation tasks to model the graph-structured relations between users and items [38,41,52].…”

Multi-Behavior Enhanced Recommendation with Cross-Interaction Collaborative Relation Modeling

“…However, these recommendation solutions mostly focus on singular type of user-item interaction behavior. In real-world online applications, users' behaviors are multi-typed in nature, *Corresponding author: Yong Xu. which involves heterogeneous relations (e.g., browse, rating, purchase) between user and item [3], [16]. Each type of user behavior may exhibit different semantic information for characterizing diversified interactive patterns between user and item.…”

“…Such different types of interaction behaviors may be correlated in complex ways to provide complementary information for learning user interests. Additionally, although there exist some recent developed multi-behavior user modeling techniques for recommendation [2], [16], they fail to capture the high-order collaborative effects with the awareness of different user-item relations. Taking the inspiration from the effectiveness by employing graph neural networks in recommendation [11], [13], it is beneficial to consider high-order relations between user-item interaction into the embedding space.…”