GSLB

GSLB是由中国科学院自动化研究所开发的综合性图结构学习基准，包含16种先进的图结构学习算法和20个多样的图数据集，覆盖了同质节点级、异质节点级和图级任务。该数据集旨在通过三个维度：有效性、鲁棒性和复杂性，系统地研究图结构学习的特性。GSLB不仅为图结构学习领域提供了标准化的实验设置，还通过广泛的实验揭示了图结构学习在异质性、鲁棒性等方面的潜在优势，为未来的研究提供了深入的见解和机会。

GSLB is a comprehensive graph structure learning benchmark developed by the Institute of Automation, Chinese Academy of Sciences. It encompasses 16 state-of-the-art graph structure learning algorithms and 20 diverse graph datasets, covering homogeneous node-level, heterogeneous node-level, and graph-level tasks. This benchmark aims to systematically investigate the characteristics of graph structure learning across three dimensions: effectiveness, robustness, and complexity. Not only does GSLB provide standardized experimental setups for the graph structure learning community, but it also reveals the potential advantages of graph structure learning in aspects such as heterogeneity and robustness through extensive experiments, offering in-depth insights and opportunities for future research.