Training dataset for finetuned indobert

This data consist from the top 5 stock investing application in Indonesia, I select the apps that only have more than 1 million downloads in Google Play Store. The source of this data solely from Google Play Store scraped with Google-play-scraper library in python. This data was taken with purposive sampling method with these rules:- Only data from 1 January 2024 until 15 January 2025.- Only reviews with more than three words.- We take the smallest number of reviews from all apps, so if App X have 300 reviews, App Y 250 reviews, App Z 232 reviews. We take the last 232 reviews from all application to ensure fairness with the number of reviews taken from all apps.本数据集包含印度尼西亚排名前五的股票投资类应用程序，本次研究仅选取谷歌应用商店（Google Play Store）下载量超100万次的应用作为样本。本数据集的全部数据均通过Python的google-play-scraper库从谷歌应用商店爬取获得。本次数据采集采用目的性抽样（purposive sampling）法，遵循以下规则：- 仅采集2024年1月1日至2025年1月15日期间的评论数据；- 仅保留字数超过3的评论；- 以所有应用中评论数最少的样本量作为统一采样标准以保障公平性：例如若应用X有300条评论、应用Y有250条、应用Z有232条，则从所有应用中各提取最后232条评论，确保各应用的采样规模完全一致。