Self-Attentive Sequential Recommendation with Cheap Causal Convolutions

TL;DR

This paper introduces a novel sequential recommendation model that combines self-attention with cheap causal convolutions to better utilize local context and improve sequence encoding, leading to enhanced recommendation accuracy.

Contribution

It proposes a new model integrating cheap causal convolutions with self-attention, effectively capturing local context and reducing redundancy in sequence representations.

Findings

Outperforms existing models on benchmark datasets

Improves accuracy in single- and multi-objective recommendation tasks

Reduces model complexity with lightweight convolutional structures

Abstract

Sequential Recommendation is a prominent topic in current research, which uses user behavior sequence as an input to predict future behavior. By assessing the correlation strength of historical behavior through the dot product, the model based on the self-attention mechanism can capture the long-term preference of the sequence. However, it has two limitations. On the one hand, it does not effectively utilize the items' local context information when determining the attention and creating the sequence representation. On the other hand, the convolution and linear layers often contain redundant information, which limits the ability to encode sequences. In this paper, we propose a self-attentive sequential recommendation model based on cheap causal convolution. It utilizes causal convolutions to capture items' local information for calculating attention and generating sequence embedding. It also uses cheap convolutions to improve the representations by lightweight structure. We evaluate the effectiveness of the proposed model in terms of both accurate and calibrated sequential recommendation. Experiments on benchmark datasets show that the proposed model can perform better in single- and multi-objective recommendation scenarios.