jugalgajjar/CS-Knowledge-Graph-Dataset

CS Knowledge Graph Dataset是一个基于OpenAlex构建的多尺度异构计算机科学学术知识图谱，包含论文、作者、出版场所和概念标签等节点，以及它们之间的多种关系（如作者关系、引用关系、出版关系等）。该数据集旨在支持知识图谱嵌入、链接预测、节点分类、学术推荐和图神经网络等研究任务，并提供了从1k到10m不同规模的子图，以适应从快速原型到大规模训练的需求。每个规模都是独立的子图，包含完整的节点和边信息，节点类型包括Paper、Author、Venue和Concept，边关系包括AUTHORED、CITES、PUBLISHED_IN、BELONGS_TO和COLLABORATES_WITH。数据集以Parquet格式提供，并附带原始SQLite数据库文件，方便用户进行灵活的数据处理和分析。The CS Knowledge Graph Dataset is a multi-scale heterogeneous knowledge graph of computer science scholarly data, built from OpenAlex. It includes nodes such as papers, authors, publication venues, and concept tags, along with various relationships between them (e.g., authorship, citation, publication). This dataset is designed for research on knowledge graph embeddings, link prediction, node classification, scholarly recommendation, and graph neural networks. It offers five independent subgraph scales (1k, 10k, 100k, 1m, 10m) to support tasks ranging from quick prototyping to large-scale training. Node types include Paper, Author, Venue, and Concept, while edge relations include AUTHORED, CITES, PUBLISHED_IN, BELONGS_TO, and COLLABORATES_WITH. The data is provided in Parquet format, along with raw SQLite database files for flexible processing and analysis.