ID-CoastSent: Indonesian Coastal Tourism Review Dataset

ID-CoastSent: Indonesian Coastal Tourism Review Dataset is a sentiment-annotated dataset consisting of user-generated reviews related to coastal tourism destinations in Indonesia. The data were collected from Google Maps using the Google Maps API with queries associated with beach and coastal locations across multiple regions in Indonesia.

The dataset was constructed through a multi-stage process, including data collection, cleaning, selection, and manual annotation. A total of 885,550 reviews were initially collected. After applying data cleaning criteria such as removing empty entries, duplicate records, non-Indonesian language reviews, and retaining reviews with a minimum length of 20 characters, 253,422 reviews were retained. From this subset, 21,930 reviews were selected for annotation based on rating distribution.

Each review was independently annotated by three annotators into three sentiment categories: positive, neutral, and negative. After removing 145 invalid reviews identified during the annotation process, the final annotated dataset contains 21,785 reviews. Two derived datasets are provided based on annotation agreement: a Majority Vote dataset with 21,540 reviews and an All Agree dataset with 15,832 reviews.

The dataset includes review text, cleaned text, metadata (e.g., rating, date, location identifier), and sentiment labels. It is provided in structured file formats and is intended for use in natural language processing tasks, sentiment analysis, and tourism-related data studies, particularly for Bahasa Indonesia as a low-resource language.

ID-CoastSent：印度尼西亚海岸旅游评论数据集（Indonesian Coastal Tourism Review Dataset）是一款情感标注数据集，涵盖印度尼西亚沿海旅游目的地的用户生成评论。数据通过谷歌地图API（Google Maps API）从谷歌地图（Google Maps）采集，以关联印度尼西亚多区域海滩与海岸点位的关键词进行检索。
该数据集采用多阶段流程构建，包含数据采集、清洗、筛选与人工标注四个环节。初始共采集885,550条评论，经数据清洗规则处理：移除空条目、重复记录、非印尼语评论，并保留字符长度不低于20的评论后，最终留存253,422条评论。基于评分分布，从该子集中筛选出21,930条评论用于标注工作。
每条评论由三名标注人员独立标注为正面、中性、负面三类情感类别。剔除标注过程中识别出的145条无效评论后，最终标注数据集共包含21,785条评论。基于标注一致性规则，衍生出两款子数据集：包含21,540条评论的多数投票数据集（Majority Vote dataset），以及包含15,832条评论的完全一致数据集（All Agree dataset）。
该数据集包含评论原文、清洗后文本、元数据（如评分、发布日期、位置标识符）与情感标签，以结构化文件格式提供，可应用于自然语言处理、情感分析及旅游相关数据研究，尤其适配作为低资源语言的印尼语（Bahasa Indonesia）相关任务。