DBLP Discovery Dataset (D3)

DBLP Discovery Dataset (D3)是由乌珀塔尔大学德国和加拿大国家研究委员会创建的大型学术元数据集，包含超过639万篇计算机科学领域的出版物。数据集通过从DBLP获取超过600万篇出版物，并从全文提取相关元数据（如摘要、作者隶属关系、引用）来构建。D3可用于识别研究活动、生产力、焦点、偏见、可访问性和计算机科学研究影响的趋势。数据集内容丰富，包括出版物ID、标题、年份、类型、访问链接、DOI、出版商、作者ID、全名、网页、场地名称、缩写等。D3的创建过程涉及从DBLP提取所有记录，并从关联的全文PDF中提取元数据。数据集的应用领域广泛，旨在解决计算机科学研究中的各种问题，如研究增长趋势、作者活跃度、特定领域的流行话题等。

The DBLP Discovery Dataset (D3) is a large-scale academic metadata dataset jointly developed by the University of Wuppertal (Germany) and the National Research Council Canada. It contains over 6.39 million computer science publications. The dataset is constructed by retrieving more than 6 million publications from DBLP and extracting relevant metadata including abstracts, author affiliations, and citations from their full texts. D3 can be utilized to identify trends related to research activities, productivity, research focuses, biases, accessibility, and the impact of computer science research. The dataset covers rich metadata fields such as publication ID, title, year, type, access links, DOI, publisher, author ID, full name, webpage, venue name, and abbreviation. The development process of D3 involves extracting all records from DBLP and harvesting metadata from the associated full-text PDF files. The dataset has a broad range of application scenarios, aiming to address diverse issues in computer science research, such as research growth trends, author activity levels, popular topics in specific domains, and more.