Hearst Publications Lifestyle & Culture AI Training Bundle
收藏资源简介：
This dataset is a **high-volume, historically significant text corpus** that excels as raw training material for language models focused on period-authentic American English. Its greatest strength is the sheer breadth of content, 700K+ articles from six iconic publications spanning the most transformative period in American media history. The rigorous OCR work has produced a structurally complete, zero-gap text field that is immediately usable for pre-training and fine-tuning. This is not a collection of random articles. This is the written record of American domestic life, fashion, literature, and culture from the Gilded Age through the Jazz Age, six iconic magazines spanning 63 years (1867-1930). Cleaned. Structured. Bias-audited. Six publications. 704,930 rows. 63 years of cultural history. <p><br/></p> **Included publications:** - Harper's Bazaar (1867-1930) – High fashion, art, literature, beauty. The authority on style for over six decades. - Cosmopolitan (1886-1930) – Fiction, social issues, women's lifestyle. From the Gilded Age to the Jazz Age. - Good Housekeeping (1885-1930) – Home economics, interior design, family health, product testing. The trusted voice of the American home. - Redbook (1903-1930) – Fiction, marriage, parenting, home management. Literary and cultural commentary. - House Beautiful (1896-1930) – Interior design, architecture, home decoration. The guide to beautiful living. - The Connoisseur (1905-1930) – Fine arts, antiques, collecting, and art criticism. The magazine for art lovers and collectors. <p><br/></p> All content is professionally OCR-cleaned, structured as JSONL, and bias-audited with historical context notices. 100% public domain (pre-1930). Snowflake-optimized. <p><br/></p> # 3 Sample Cortex Agent Prompts: Use these prompts directly with this dataset to explore its capabilities: **Prompt 1 — Cultural Content Discovery:** > *"Find articles from Harper's Bazaar between 1890 and 1910 that discuss fashion or dress styles. Show me the titles and issue references."* **Prompt 2 — Publication Comparison:** > *"Count the number of articles by publication and show me how the content volume is distributed across Harper's Bazaar, Cosmopolitan, Good Housekeeping, Redbook, and House Beautiful."* **Prompt 3 — Historical Recipe & Domestic Life Research:** > *"Search Good Housekeeping for content mentioning recipes, kitchen, or cooking. Return 20 sample titles with their issue identifiers so I can explore domestic life content from this era."* <p><br/></p> Total rows: 704,930 Time period: 1867-1930 Publications: 6 This is a Devin Media Corp exclusive.
数据集概述：Hearst Publications Lifestyle & Culture AI Training Bundle
发布方：Devin Media Corp.
数据集简介： 该数据集收录了六本美国标志性杂志在1867年至1930年（共63年）间的历史内容，总计704,930行。内容涵盖时尚、家居、机械、文学、文化等多个领域，所有内容均来自公共领域（1930年之前），已进行专业的OCR清洗、结构化处理（JSONL格式）和偏见审计，并针对Snowflake平台进行了优化。
收录出版物：
- Harpers Bazaar (1867-1930)：高级时尚、艺术、文学、美容。
- Cosmopolitan (1886-1930)：小说、社会议题、女性生活方式。
- Good Housekeeping (1885-1930)：家政学、室内设计、家庭健康、产品测试。
- Redbook (1903-1930)：小说、婚姻、育儿、家庭管理。
- House Beautiful (1896-1930)：室内设计、建筑、家居装饰。
- The Connoisseur (1905-1930)：美术、古董、收藏、艺术评论。
关键规模：
- 总行数：704,930
- 时间跨度：1867-1930
- 出版物数量：6
数据用途：
- 训练时尚AI模型（基于Harpers Bazaar 60多年的时尚版式、描述和趋势）。
- 分析家政学、室内设计和消费品测试的演变。
- 训练文学AI（包括Cosmopolitan和Redbook中的美国主要作家作品）。
- 研究广告文案、品牌语言和消费者心理的演变。
- 支持物质文化、性别研究和家庭生活史研究。
数据结构（数据字典）：
数据表名为 HEARST_CULTURE，包含以下列：
ISSUE(Varchar)：期刊期号标识。TITLE(Varchar)：文章标题。AUTHOR(Varchar)：作者（部分未知）。TYPE(Varchar)：内容类型（例如"article"）。TEXT(Varchar)：文章正文内容。INGESTION_DATE(Timestamp_NTZ)：数据摄入时间。
业务需求：
- 模型开发：支持生活方式和文化AI的预训练和微调，包括时尚AI、家居生活应用、文学分析及文化趋势建模。
- 实体识别：提取时尚实体（服装、面料、配饰、设计师）、家政学概念（产品测试、室内设计、食谱）及文学主题，适用于19世纪和20世纪初文化术语的高精度命名实体识别（NER）。
- 文本摘要：提供全文文章与原始标题的配对，用于训练生成简洁摘要的模型。
使用示例（SQL查询）：
- 查看数据集内容：
SELECT ISSUE, TITLE, AUTHOR FROM HEARST_CULTURE WHERE TYPE = article LIMIT 10; - 查找时尚内容（Harpers Bazaar）：根据
ISSUE和TEXT中的关键词（如fashion, dress, bonnet）进行筛选。 - 查找家政学内容（Good Housekeeping）：根据
ISSUE和TEXT中的关键词（如recipe, home, kitchen）进行筛选。 - 按出版物统计文章数：使用
CASE WHEN语句根据ISSUE分组并进行COUNT(*)统计。
定价与访问：
- 定价：需点击“Get Unlock New Insights”获取详细信息。
- 交付方式：Secure share（安全共享）。
- 刷新频率：每年。
其他数据集（来自Devin Media Corp.）：
- New York Journal of Medicine Archive (1843-1860)
- Meteorological Magazine (1919–1930)
- JAMA Pediatrics Clinical Dataset (1911–1930)
- ICIS Chemical Business Archive (1876–1930)
- Journal of the American Medical Association (Dermatology) — Complete Archive 1920-1930
- JAMA Ophthalmology (1869 - 1930) Trial
