---
title: "GraphSculptor: Sculpting Pre-training Core Sets for Graph Self-supervised Learning"
authors:
  - "Chuang Liu"
  - "Zelin Yao"
  - "Xueqi Ma"
  - "Luzhi Wang"
  - "Mukun Chen"
  - "Pinghua Xu"
  - "Wenbin Hu"
doi: "10.24963/ijcai.2026/316"
url: "https://ijcai.org/proceedings/2026/316"
pdf: "https://ijcai.org/proceedings/2026/0316.pdf"
venue: "Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence"
year: 2026
pages: "2843-2851"
track: "Main Track"
keywords:
  - "Data Mining: Mining graphs"
---

# GraphSculptor: Sculpting Pre-training Core Sets for Graph Self-supervised Learning

Chuang Liu, Zelin Yao, Xueqi Ma, Luzhi Wang, Mukun Chen, Pinghua Xu, Wenbin Hu

*Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence*, pages 2843-2851. DOI: [10.24963/ijcai.2026/316](https://doi.org/10.24963/ijcai.2026/316)

## Abstract

Graph self-supervised learning (SSL) typically relies on large-scale unlabeled datasets, heavily inflating computational costs. However, empirical evidence suggests that these datasets contain substantial redundancy—our analysis reveals that uniformly subsampling 50% of graphs retains over 96% of downstream performance. To exploit this redundancy, we introduce GraphSculptor for pre-training coreset construction. Unlike methods dependent on additional training-time signals or limited solely to topological statistics, GraphSculptor provides a label-free solution that constructs coresets via two complementary perspectives: intrinsic structure and contextual semantics. Concretely, structural diversity is quantified using intrinsic graph statistics, yielding a structural feature vector for each graph, while semantic diversity is captured by utilizing a pre-trained language model to encode descriptions generated via graph-to-text. GraphSculptor integrates these signals into a unified metric space and performs cluster-aware selection to preserve joint structural--semantic diversity. We further derive a theoretical bound on the loss gap between coreset and full-data pre-training, offering theoretical motivation for our selection formulation. Extensive experiments demonstrate that GraphSculptor effectively "sculpts" the dataset: a 10% coreset achieves 99.6% of full-data performance while reducing pre-training time by nearly 90%, offering a scalable solution for data-efficient graph pre-training.