biglam/AmericanStories-parquet

AmericanStories (parquet) 是一个大型英语历史报纸文章数据集，涵盖从1774年到1963年的广泛时间范围，按年份组织成多个配置。数据集源自dell-research-harvard/AmericanStories，以parquet格式存储，规模在1000万到1亿条目之间。特征包括文章ID、报纸名称、版次、日期、页码、标题、署名和文章内容，适用于文本分类、文本生成、文本检索、摘要和问答等多种NLP任务。标签表明数据集涉及社会科学、经济学、新闻、报纸、大语言模型（LLM）、NLP、历史研究等领域，并关联到arXiv论文2308.12477。数据集主要用于历史语言建模、社会经济学分析和新闻文本处理研究。

AmericanStories (parquet) is a large-scale English historical newspaper article dataset spanning a wide time range from 1774 to 1963, organized into multiple configurations by year. The dataset is derived from dell-research-harvard/AmericanStories, stored in parquet format, with a size between 10 million and 100 million entries. Features include article ID, newspaper name, edition, date, page, headline, byline, and article content, suitable for various NLP tasks such as text classification, text generation, text retrieval, summarization, and question answering. Tags indicate the dataset covers social science, economics, news, newspaper, large language modeling (LLM), NLP, historical research, and is linked to arXiv paper 2308.12477. The dataset is primarily used for historical language modeling, socio-economic analysis, and news text processing research.