IntelliGraphs

IntelliGraphs数据集由阿姆斯特丹大学等机构创建，包含五个专门设计用于评估子图推理模型的知识图谱数据集。这些数据集包括合成数据和从Wikidata提取的实际数据，旨在测试模型在生成逻辑上有效且新颖的子图方面的能力。数据集的创建过程中，使用了逻辑规则来确保子图的语义一致性，并开发了一个Python包来随机生成和验证子图。IntelliGraphs的应用领域主要集中在推动机器学习模型在理解知识图谱中的语义关系方面的发展，特别是在需要逻辑推理和语义理解的场景中。The IntelliGraphs dataset, created by institutions including the University of Amsterdam, consists of five knowledge graph datasets specifically designed to evaluate subgraph reasoning models. These datasets cover both synthetic data and real-world data extracted from Wikidata, and are intended to test models' capability to generate logically valid and novel subgraphs. During the construction of the dataset, logical rules were adopted to ensure the semantic consistency of subgraphs, and a Python package was developed for randomly generating and validating subgraphs. The application scenarios of IntelliGraphs mainly focus on advancing the development of machine learning models in understanding semantic relationships within knowledge graphs, particularly in scenarios requiring logical reasoning and semantic comprehension.