Session recommendation gets faster and better without attention models
No Attention, No Problem: Rethinking Session-based Recommendation with Pure Convolution
Information Retrieval
Summary
Session-based recommendation systems try to predict what you will choose next based on your recent actions. Usually, complex models that use attention mechanisms help these systems understand long-term connections but are slower. The authors show that a new convolution-based method can do this faster while keeping or even improving accuracy. This method combines positional signals and structural information without using attention, making it more efficient for real-time predictions.
What this means in practice
- •For real-time recommendation engineers: Speed up next-item predictions in session recommender systems while improving accuracy by avoiding slow attention mechanisms
- •For e-commerce platform developers: Deploy efficient recommendation models that handle long user sessions without expensive attention layers, optimizing costs and user experience
Authors
Tao Huang, Wei Zhou
Abstract
Session-based recommendation (SBR) predicts the next choice in a session by analyzing recent interactions. Transformer-based models are widely used because of their ability to capture long-range dependencies through self-attention mechanisms. In contrast, traditional convolutional models, although more efficient, are often limited by their weak global modeling capabilities and are losing ground in SBR tasks. In this work, we propose a Next-generation Pure Convolutional Framework (NextConvRec) for SBR tasks, aiming to balance efficiency and performance. NextConvRec uses a Structural and Positional Convolutional Encoder (SPCE) for preprocessing, combining learnable convolutional positional biases with session-level structural signals extracted through GCN layers. Its backbone convolutional module effectively expands the effective receptive field through depthwise convolutions and pointwise convolutions, enabling robust long-range preference modeling without attention mechanisms. Extensive experiments on 4 benchmark datasets show that NextConvRec outperforms several state-of-the-art baselines by around 1.73% on average, and reduces the average inference time per session by 16.7%. The convolutional architectures remain a promising direction for efficient and accurate session-based recommendations.