tlg-dataset

tlg-dataset是由悉尼大学创建的一个包含18,793篇新闻文章的数据集，涵盖39个实体。该数据集旨在通过选择关键事件的故事来总结实体的生平，解决时间线生成任务中缺乏标准数据集和评估方法的问题。数据集内容丰富，包括每个实体的金标准时间线和相关文章，以及用于进一步开发的主题链接文档。创建过程涉及从Wikipedia中提取文章进行注释，以生成金标准。该数据集适用于时间线生成任务的评估和模型开发，旨在提供未来研究的标准基准。

The TLG-Dataset is a dataset containing 18,793 news articles covering 39 entities, created by the University of Sydney. This dataset aims to summarize entities' life stories by selecting narratives of key events, addressing the lack of standard datasets and evaluation methods for timeline generation tasks. The dataset comprises rich content, including gold-standard timelines and related articles for each entity, alongside topic-linked documents for further development. Its creation process involves extracting articles from Wikipedia for annotation to generate the gold standards. This dataset is applicable to the evaluation of timeline generation tasks and model development, and aims to provide a standard benchmark for future research.