nlp-waseda/e_gov

这是一个从e-Gov获取的日本法律数据集。数据集中的每条数据包含两个字段：text字段包含法律文本，metadata字段包含包括以下9个子字段的额外信息：Era（法律颁布的日本时代），Lang（文本所用的语言，均为日语），LawType（法律类型，包括宪法、法案、内阁命令、敕令、省令、规则和其他），Year（法律颁布年份），PromulgateMonth/Day（法律颁布的月份/日期），LawNum（法律的数字名称），category_id（表示法律分类的整数）。数据集被随机分为训练集、验证集和测试集，比例为8:1:1，保持原始分类分布。

This is a Japanese law dataset obtained from e-Gov. Each entry in the dataset contains two fields: text field with legal texts, and metadata field with additional information including nine subfields: Era (the Japanese Era when the law is promulgated), Lang (the language of the text, all in Japanese), LawType (the type of the law, including Constitution, Act, CabinetOrder, ImperialOrder, MinisterialOrdinance, Rule, and Misc), Year (the year when the law is promulgated), PromulgateMonth/Day (the month/day when the law is promulgated), LawNum (the numeric name of the law), and category_id (an integer representing the category of the law). The dataset is split into train, validation, and test sets at a ratio of 80%:10%:10%, preserving the original distribution of categories.