TopKGAT: A Top-K Objective-Driven Architecture for Recommendation

TL;DR

TopKGAT is a new recommendation architecture designed to optimize directly for top-K metrics like Precision@K, leading to improved accuracy in recommending relevant items, and is based on a differentiable approximation of these metrics.

Contribution

It introduces a novel architecture that aligns model training with top-K objectives using a differentiable approximation, enhancing recommendation effectiveness.

Findings

TopKGAT outperforms state-of-the-art baselines on four benchmark datasets.

The architecture is efficient and resembles a graph attention network.

It directly optimizes for top-K metrics, improving recommendation relevance.

Abstract

Recommendation systems (RS) aim to retrieve the top-K items most relevant to users, with metrics such as Precision@K and Recall@K commonly used to assess effectiveness. The architecture of an RS model acts as an inductive bias, shaping the patterns the model is inclined to learn. In recent years, numerous recommendation architectures have emerged, spanning traditional matrix factorization, deep neural networks, and graph neural networks. However, their designs are often not explicitly aligned with the top-K objective, thereby limiting their effectiveness. To address this limitation, we propose TopKGAT, a novel recommendation architecture directly derived from a differentiable approximation of top-K metrics. The forward computation of a single TopKGAT layer is intrinsically aligned with the gradient ascent dynamics of the Precision@K metric, enabling the model to naturally improve top-K recommendation accuracy. Structurally, TopKGAT resembles a graph attention network and can be implemented efficiently. Extensive experiments on four benchmark datasets demonstrate that TopKGAT consistently outperforms state-of-the-art baselines. The code is available at https://github.com/StupidThree/TopKGAT.