Companies / Other
National Institute of Korean Language
Korean government institute that commissioned the national written corpus for AI, built from 20,053 works of which only part were licensed from their publishers. Recorded here: 2 public prices.
South Koreakorean.go.krLatest record 13 Sep 2022
Public prices involving National Institute of Korean Language
- Per-title settlement Woongjin Booxen set with the publishers whose Bookopia e-books it used without permission in the National Institute of Korean Language written corpus (an AI and language-research corpus), Korea, 2022
A formula, not an amount: about 2.1 times the e-book list price per title. Woongjin called it a settlement of fees, not compensation. Seoul Shinmun (26 August 2022) reports 6,194 of the 15,993 disputed titles were settled at the level paid to publishers under formal contracts.
Press report, Kyunghyang Shinmun, opinion column by culture researcher Lee Yung-hee, 13 September 2022
전자책 정가의 70%의 3copy (70% of the e-book list price, times 3 copies)per title, as a multiple of the e-book list price
- Korean government (National Institute of Korean Language) written-corpus collection project: total project cost over the 20,053 published works the contractor used from 1,226 publishers, 2019
A buyer paying a vendor to assemble a corpus, so the cost covers processing as well as rights; only 4,060 works were licensed from their publishers, and 15,933 Bookopia e-books were used without consent (later settled, see the related Korean rows). Bizhankook reports the contract required at least 80% of the fee to go to copyright clearance.
Press report, Seoul Shinmun, Woongjin Booxen infringed copyright in 16,000 titles (26 August 2022)
총 사업비는 30억 9000만원 (3.09 billion won) for 2만 53종 (20,053 works)per project of 20,053 works
Cite this
AIData.Broker, “National Institute of Korean Language in the AI training data market”, reviewed 13 Sep 2022. https://aidata.markets/companies/national-institute-of-korean-language
Free to reuse under CC BY 4.0 if you credit AIData.Broker and link to this page. Each row names the publication that stated it, and words quoted from a source keep that source's own terms.