cmotions/NL_restaurant_reviews

---
language:
- nl
tags:
- text-classification
- sentiment-analysis
datasets:
- train
- test
- validation
---

## Dataset overview
This is a dataset that contains restaurant reviews gathered in 2019 using a webscraping tool in Python. Reviews on restaurant visits and restaurant features were collected for Dutch restaurants.
The dataset is formatted using the 🤗[DatasetDict](https://huggingface.co/docs/datasets/index) format and contains the following indices:
- train, 116693 records
- test, 14587 records
- validation, 14587 records

The dataset holds both information of the restaurant level as well as the review level and contains the following features:
- [restaurant_ID] > unique restaurant ID
- [restaurant_review_ID] > unique review ID
- [michelin_label] > indicator whether this restaurant was awarded one (or more) Michelin stars prior to 2020
- [score_total] > restaurant level total score
- [score_food] > restaurant level food score
- [score_service] > restaurant level service score
- [score_decor] > restaurant level decor score
- [fame_reviewer] > label for how often a reviewer has posted a restaurant review
- [reviewscore_food] > review level food score
- [reviewscore_service] > review level service score
- [reviewscore_ambiance] > review level ambiance score
- [reviewscore_waiting] > review level waiting score
- [reviewscore_value] > review level value for money score
- [reviewscore_noise] > review level noise score
- [review_text] > the full review that was written by the reviewer for this restaurant
- [review_length] > total length of the review (tokens)

## Purpose
The restaurant reviews submitted by visitor can be used to model the restaurant scores (food, ambiance etc) or used to model Michelin star holders. In [this blog series](https://medium.com/broadhorizon-cmotions/natural-language-processing-for-predictive-purposes-with-r-cb65f009c12b) we used the review texts to predict next Michelin star restaurants, using R.

原始信息汇总

数据集概述

该数据集包含2019年通过Python网络爬虫工具收集的荷兰餐厅评论。这些评论涉及餐厅访问和餐厅特色。数据集采用🤗DatasetDict格式，包含以下部分：

• 训练集：116693条记录
• 测试集：14587条记录
• 验证集：14587条记录

数据集特征

数据集包含餐厅级别和评论级别的信息，具体特征如下：

• restaurant_ID：唯一餐厅ID
• restaurant_review_ID：唯一评论ID
• michelin_label：2020年前该餐厅是否获得米其林星级
• score_total：餐厅总评分
• score_food：餐厅食品评分
• score_service：餐厅服务评分
• score_decor：餐厅装饰评分
• fame_reviewer：评论者发布评论的频率
• reviewscore_food：评论食品评分
• reviewscore_service：评论服务评分
• reviewscore_ambiance：评论氛围评分
• reviewscore_waiting：评论等待评分
• reviewscore_value：评论性价比评分
• reviewscore_noise：评论噪音评分
• review_text：评论全文
• review_length：评论长度（词数）

数据集用途

该数据集可用于模型化餐厅评分（如食品、氛围等）或预测米其林星级餐厅。在此博客系列中，我们使用评论文本预测下一个米其林星级餐厅，使用R语言。