Beyond Clicks: Modeling Multi-Relational Item Graph for Session-Based Target Behavior Prediction

Abstract

Session-based target behavior prediction aims to predict the next item to be interacted with specific behavior types (e.g., clicking). Although existing methods for session-based behavior prediction leverage powerful representation learning approaches to encode items' sequential relevance in a low-dimensional space, they suffer from several limitations. Firstly, they focus on only utilizing the same type of user behavior for prediction, but ignore the potential of taking other behavior data as auxiliary information. This is particularly crucial when the target behavior is sparse but important (e.g., buying or sharing an item). Secondly, item-to-item relations are modeled separately and locally in one behavior sequence, and they lack a principled way to globally encode these relations more effectively. To overcome these limitations, we propose a novel Multirelational Graph Neural Network model for Session-based target behavior Prediction, namely MGNN-SPred for short. Specifically, we build a Multi-Relational Item Graph (MRIG) based on all behavior sequences from all sessions, involving target and auxiliary behavior types. Based on MRIG, MGNN-SPred learns global itemto-item relations and further obtains user preferences w.r.t. current target and auxiliary behavior sequences, respectively. In the end, MGNN-SPred leverages a gating mechanism to adaptively fuse user representations for predicting next item interacted with target behavior. The extensive experiments on two real-world datasets demonstrate the superiority of MGNN-SPred by comparing with state-of-the-art session-based prediction methods, validating the benefits of leveraging auxiliary behavior and learning item-to-item relations over MRIG.

Exploiting Cross-session Information for Session-based Recommendation with Graph Neural Networks

“…We will discuss the relationship and difference between the proposed PosRec model and existing recommendation methods as well as the position encoding schemes. , FGNN [29,30], and MGNN-SPred [47], PosRec has a newly designed GNN module, PGGNN, which is position aware. The position awareness is the main difference between the GNN modules.…”

“…These attention-based methods only consider the importance of the last position while neglecting other positions. (3) In graph modeling, e.g., SR-GNN [49], GC-SAN [51], FGNN [30] and MGNN-SPred [47], a session is converted into a graph and Graph Neural Networks (GNN) [19,25,43] captures the connectivity of items. Afterward, a readout function is applied to compute a session representation with the processed item representations.…”

“…Such a strategy for the positional information can only consider the positional information of the last interaction, and neglect the positional information of all other interactions. While for the GNN-based methods [29,30,47,49,51], the relative positional information is contained in the direction of edges. But the relative positional information cannot reflect the absolute positional information, e.g., the latest preference and the initial intention.…”

Graph Learning based Recommender Systems: A Review

“…[10] [Qiu et al, 2019]; [11] [Wang et al, 2020c]; [12] ]; [13] [Wang et al, 2019f]; [14] [Jamali and Ester, 2009]; [15] [Wen et al, 2018]; [16] [Fan et al, 2019]; [17] [Gao et al, 2019]; [18] [ Table 1: A summary of data in RS, the representing graph, and the corresponding GLRS approach ing a given time period, and ordered by their time stamp. According to the number of interaction types included in a sequence, a sequential interaction data set can be divided into single-type interaction data set where only one type of interactions is included, and multi-type interaction data set where multiple types of interactions are included.…”

Context-Aware Short-Term Interest First Model for Session-Based Recommendation

“… Pop is a simple baseline that recommends top rank items based on popularity in training data.  Item-KNN [30] recommends the item similar to the items that have been clicked in the current session, where the cosine similarity is used.  BPR-MF [28] is a learning-to-rank method.…”