DBLP Temporal

数字书目和图书馆项目 (DBLP) 是由特里尔大学开发和维护的计算机科学书目数据库，在过去的三十年里，许多研究人员都将其用于实验研究。到 2018 年，DBLP 包含超过 410 万篇出版物。虽然 DBLP 试图识别单个作者，但在许多情况下，由同音异义词或同名引起的错误不会被检测到。 DBLP 有可能被用作以出版物为记录的时间数据集，每个记录都与时间戳（即出版日期）相关联，并且作者由其唯一标识符标识。由于作者可能会随着时间的推移而改变他们的从属关系，因此识别在其整个职业生涯中提及同一作者的所有记录具有挑战性。The Digital Bibliography & Library Project (DBLP) is a computer science bibliography database developed and maintained by the University of Trier. It has been widely adopted by researchers for experimental studies over the past three decades. As of 2018, DBLP contained over 4.1 million publications. Although DBLP aims to identify individual authors, errors induced by homonyms or identical names often remain undetected in numerous cases. DBLP can be employed as a publication-centric temporal dataset, where each record is linked to a timestamp (i.e., publication date) and authors are identified via their unique identifiers. Given that authors may alter their affiliations over time, identifying all records referring to the same author across their entire career presents a considerable challenge.

搜集汇总

背景与挑战

背景概述

DBLP Temporal是一个基于DBLP计算机科学书目数据库的时间数据集，包含超过410万篇出版物，重点关注作者实体解析问题，由于同名或音似以及作者从属关系随时间变化，数据集适用于动态链接预测等研究任务。该数据集由澳大利亚国立大学于2018年发布，提供相关论文和代码资源。