Myanmar Sentiment Corpus

 A sentiment classification corpus of 2,999 public product-review comments in Burmese (Myanmar language), collected from public Facebook groups and YouTube posts (March–May 2025), restricted to low-sensitivity product-review domains. First Myanmar sentiment corpus with documented inter-annotator agreement (Fleiss' κ = 0.652, three native-speaker annotators, majority-vote labels; per-annotator labels included). No author identities, usernames, timestamps, or URLs were retained; PII was removed before annotation.
Access is restricted: granted to researchers for non-commercial research use only; redistribution is not permitted. A takedown policy applies — any comment's author may request removal via the contact address. An ethics-determination application covering collection, continued use, and sharing is filed with the Chiang Mai University Research Ethics Committee (project no. 2569/405). See the README and the accompanying paper (under review at Array) for full governance details.