---
title: "MetaSieve：基于 SQL 元路径选择加速关系深度学习"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2608.25903v1"
permalink: "https://feed.astralor.com/items/itm_39ea06aff2748c98b58123db"
published: "2026-08-26T15:22:52.000Z"
score: 71
channel: "ai"
tags: ["MetaSieve","RelBench","SQL","关系深度学习","图神经网络"]
---

# MetaSieve：基于 SQL 元路径选择加速关系深度学习

## AI 点评

> 关系数据库上的 GNN 训练常常被子图采样拖慢，MetaSieve 把元路径选择交给 SQL 统计，工程味道很足。对同时做数据库和图学习的研究者，以及需要处理多表数据的算法团队来说，都值得一读。它不碰 GNN 参数，意味着可以低成本地试到现有框架里。

## 速览

关系深度学习将多表数据库建模为图并训练图神经网络，但子图采样成本高昂。MetaSieve 利用 SQL 的 join 与聚合能力，在元路径扩展时计算统计量，通过新颖评分函数剪枝不具信息量的元路径，且不依赖 GNN 参数，可适配多种骨干网络。在 RelBench 基准上，MetaSieve 大幅降低每轮训练时间，同时保持甚至提升准确率。

## 关键点

- MetaSieve 利用 SQL join 与聚合查询统计元路径的有用性，剪枝低信息量路径，降低子图规模。
- 元路径选择只依赖数据库统计量和任务标签，不依赖 GNN 参数，易集成到不同 GNN 架构。
- 在 RelBench 基准上，MetaSieve 显著减少每轮训练时间，且准确率通常不降反升。
- 该方法为多表关系数据库上的 GNN 训练提供了一条高效且通用的加速路径。

## 全文

Relational Deep Learning (RDL) is an effective approach to machine learning over multi-table relational databases. In RDL, a database is modeled as a graph in which each row is a node and each foreign-key relation is an edge, and a graph neural network (GNN) is trained on this graph. Training a GNN requires sampling a subgraph around every seed node in the training set, and the cost of training is largely determined by the size of these subgraphs. This paper aims to reduce subgraph size by leveraging the join and aggregation capabilities of relational database systems. We observe that sampled subgraphs are obtained by following metapaths composed of foreign-key links, and that many of these metapaths can be pruned without loss of accuracy. We present MetaSieve, a metapath selection layer that determines which metapaths to retain and which to prune. For each candidate metapath extension, MetaSieve computes statistics via SQL join and aggregation queries and evaluates the extension based on a novel scoring function that prefers lightweight but informative candidates. Metapaths whose scores fall below a threshold are deemed uninformative and pruned. Metapath selection in MetaSieve is lightweight since it relies only on database statistics and task labels, and it is independent of GNN parameters, so it integrates with diverse GNN architectures for classification and regression. Our evaluation on the RelBench benchmark with multiple GNN backbones shows that MetaSieve consistently reduces per-epoch training time by large margins while maintaining and often improving accuracy.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_39ea06aff2748c98b58123db