alpindale/two-million-bluesky-posts

该数据集包含从Bluesky Social的firehose API收集的200万条公开帖子，旨在用于机器学习和社交媒体数据的研究。每条帖子包含文本内容、元数据、媒体附件信息以及回复关系。数据集有两种配置：默认配置和带有语言预测的配置。默认配置包含帖子的文本内容、创建时间、作者、唯一标识符、是否包含图片以及回复关系等信息。带有语言预测的配置在默认配置的基础上增加了预测语言和语言置信度。数据集的使用受Bluesky的服务条款约束，并且主要用于训练和测试语言模型、分析社交媒体发布模式、研究对话结构和回复网络、以及社交媒体内容审核研究等。

This dataset contains 2 million public posts collected from Bluesky Socials firehose API, intended for machine learning research and experimentation with social media data. The dataset is available in two configurations: the default configuration includes fields such as text content, creation timestamp, author, unique identifier, presence of images, and reply relationships. The other configuration adds language predictions using the glotlid model, providing language codes and confidence scores. The dataset can be used for training and testing language models, analyzing social media posting patterns, studying conversation structures and reply networks, content moderation research, and natural language processing tasks.