m3hrdadfi/recipe_nlg_lite
收藏资源简介:
RecipeNLG是一个用于半结构化文本生成的烹饪食谱数据集,包含7,198个烹饪食谱。该数据集经过精心处理,提供了比该领域其他数据集更多的样本。数据集的结构包括食谱的唯一标识符(uid)、名称(name)、描述(description)、链接(link)、命名实体识别(ner)、配料(ingredients)和步骤(steps)。
RecipeNLG is a culinary recipe dataset tailored for semi-structured text generation, encompassing 7,198 cooking recipes. This dataset has been meticulously processed, offering a larger volume of samples relative to other datasets within this research domain. The structure of the dataset includes the following core fields for each recipe: unique identifier (uid), name, description, link, named entity recognition (NER), ingredients, and steps.
RecipeNLG数据集概述
数据集基本信息
- 名称: RecipeNLG
- 版本: Lite
- 描述: 用于半结构化文本生成的烹饪食谱数据集
- 规模: 包含7,198个烹饪食谱
数据集结构
- 数据集类型: DatasetDict
- 子集:
- 训练集: 包含6118个样本
- 测试集: 包含1080个样本
- 特征:
uid: 唯一标识符name: 食谱名称description: 食谱描述link: 食谱链接ner: 命名实体识别ingredients: 食材列表steps: 制作步骤
示例数据
json { "description": "...", "ingredients": "...", "link": "...", "name": "...", "ner": "...", "steps": "...", "uid": "..." }
引用信息
bibtex @misc{RecipeNLGLite, author = {Mehrdad Farahani}, title = {RecipeNLG: A Cooking Recipes Dataset for Semi-Structured Text Generation (Lite)}, year = 2021, publisher = {GitHub}, journal = {GitHub repository}, howpublished = {url{https://github.com/m3hrdadfi/recipe-nlg-lite}} }




