GPU-accelerated Multi-relational Parallel Graph Retrieval for Web-scale Recommendations

TL;DR

This paper introduces GMP-GR, a GPU-accelerated framework for large-scale, multi-relational graph retrieval that significantly improves accuracy and efficiency in web-scale recommendation systems, exemplified by Baidu's deployment.

Contribution

The paper presents a novel multi-relational relevance metric and a hierarchical parallel graph-based ANNS method optimized for GPU, enabling scalable and effective recommendations.

Findings

GMP-GR achieves higher retrieval accuracy than existing methods.

It delivers over 100 million requests per second in Baidu applications.

The framework improves user-item relevance understanding in large-scale systems.

Abstract

Web recommendations provide personalized items from massive catalogs for users, which rely heavily on retrieval stages to trade off the effectiveness and efficiency of selecting a small relevant set from billion-scale candidates in online digital platforms. As one of the largest Chinese search engine and news feed providers, Baidu resorts to Deep Neural Network (DNN) and graph-based Approximate Nearest Neighbor Search (ANNS) algorithms for accurate relevance estimation and efficient search for relevant items. However, current retrieval at Baidu fails in comprehensive user-item relational understanding due to dissected interaction modeling, and performs inefficiently in large-scale graph-based ANNS because of suboptimal traversal navigation and the GPU computational bottleneck under high concurrency. To this end, we propose a GPU-accelerated Multi-relational Parallel Graph Retrieval (GMP-GR) framework to achieve effective yet efficient retrieval in web-scale recommendations. First, we propose a multi-relational user-item relevance metric learning method that unifies diverse user behaviors through multi-objective optimization and employs a self-covariant loss to enhance pathfinding performance. Second, we develop a hierarchical parallel graph-based ANNS to boost graph retrieval throughput, which conducts breadth-depth-balanced searches on a large-scale item graph and cost-effectively handles irregular neural computation via adaptive aggregation on GPUs. In addition, we integrate system optimization strategies in the deployment of GMP-GR in Baidu. Extensive experiments demonstrate the superiority of GMP-GR in retrieval accuracy and efficiency. Deployed across more than twenty applications at Baidu, GMP-GR serves hundreds of millions of users with a throughput exceeding one hundred million requests per second.