RAW DATA: Consumer Sentiment toward AI-Generated Content in E-Commerce: Evidence from Indonesian Mobile Application Reviews Using IndoBERT

1. 01_raw_candidate_reviews_4827.csv The raw input of all analyses: 4,827 Indonesian-language Google Play reviews of five e-commerce applications (Shopee, Tokopedia, Lazada, Blibli, Bukalapak) retrieved by the keyword filter, with review text, star rating and the provisional keyword coding (manuscript Sections 3.3 to 3.5). For the editors and reviewers only. Not for public redistribution: the review texts belong to their authors and are subject to the terms of Google Play.2. 02_coded_reviews_4827.csv The coded data set without review text: polarity labels, final AIGC coding (Table 1, Appendix A) and the train/validation/test assignment. Suitable for public deposit; join to file 1 on review_id = reviewId.3. aigc_analysis_code.zip (separate file) Analysis scripts (Python), the Google Colab notebook used to fine-tune IndoBERT, the exported IndoBERT results (out/rerun/results.json) and the figures. Saving file 1 as data.csv in that folder and running run_all.py reproduces Tables 2 to 8 and Figures 2 to 5 of the manuscript. PROVENANCE- Source: Google Play Store, collected with the open-source google-play-scraper library (language and country set to Indonesia, newest reviews first), completed in October 2026; reviews dated 1 October 2018 to 12 October 2026.- About 500,000 reviews were collected; a keyword filter for AI- and automation-related terms retained the 4,827 candidate reviews in these files.- Reviewer names, profile images and other personal fields were not retained.