OAG-Bench
收藏资源简介:
OAG-Bench是由清华大学计算机科学与技术系开发的综合性、多方面、精细的人工策划基准,基于开放学术图谱(OAG)。该数据集覆盖了10个任务、20个数据集、70多个基线和120多个实验结果,旨在促进学术图谱挖掘的算法评估和比较。数据集内容包括从数千到数百万不等的数据量,涵盖了学术图谱的完整生命周期,包括数据预处理代码、算法实现和标准化评估协议。OAG-Bench的应用领域广泛,包括论文推荐、专家发现和学术影响力预测等,旨在解决学术数据挖掘中的关键挑战。
OAG-Bench is a comprehensive, multi-faceted, and meticulously human-curated benchmark developed by the Department of Computer Science and Technology, Tsinghua University, based on the Open Academic Graph (OAG). This benchmark covers 10 tasks, 20 datasets, over 70 baseline models, and more than 120 experimental results, aiming to facilitate algorithm evaluation and comparison for academic graph mining. It encompasses datasets with scales ranging from thousands to millions, covering the entire lifecycle of academic graphs, along with data preprocessing codes, algorithm implementations, and standardized evaluation protocols. OAG-Bench has a wide range of application scenarios, including paper recommendation, expert discovery, academic impact prediction and more, aiming to address key challenges in academic data mining.




