Acknowledgements

Thank you to these projects and their authors.

Code

• OpenKey, Tuyen Mai (GPL-3.0)
• Sparkle, Sparkle Project (MIT)

Word data

Vikey switches between English and Vietnamese using word statistics (how common each word is, and which words usually follow which) counted from these sources. The app contains only the statistics, never the text.

• FrequencyWords (OpenSubtitles 2018), Hermit Dave (CC BY-SA 4.0)
• FineWeb-2 and FineWeb, Hugging Face (ODC-By 1.0)
• HPLT v2, High Performance Language Technologies (CC0)
• MADLAD-400, Google (ODC-By 1.0)
• Reddit comments, tensorshield/reddit_dataset_157 (MIT)
• Vietnamese forums and reader comments, VnExpress, Tuổi Trẻ, Tinhte, Otofun, Dạy Nhau Học (word statistics)
• Emoji names and keywords (CLDR), Unicode, Inc. (Unicode License v3)
• Emojibase, Miles Johnson (MIT)

Test data

Used only to measure and improve accuracy; not in the app.

• ViLexNorm, Nguyen et al., EACL 2024 (CC BY-NC-SA 4.0)
• UIT-ViCTSD, UIT NLP Group (research use)
• VietMix, EACL 2026 (research use)
• Tap typing with touch sensing images, Google Research (CC BY 4.0)
• How We Type: Mobile, Aalto University (research use)
• GitHub Typo Corpus, Hagiwara & Mita, LREC 2020 (research use)
• Birkbeck spelling error corpus, Roger Mitton (free to use)
• Lists of common misspellings, Wikipedia (CC BY-SA 4.0)