Smilegate Holdings Inc.’s headquarters in Seongnam, south of Seoul. (Yonhap)
SEOUL, Jan. 21 (Korea Bizwire) — Online game developer Smilegate Holdings Inc. said Thursday that it had established datasets for hateful online comments and speech through its artificial intelligence (AI) center.
Alarmed by the sharp increase in hate expressions in online spaces, Smilegate’s AI center established datasets for hateful online comments and speech as a way to preemptively detect and respond to them.
Smilegate collected a large number of data samples from portal websites and online communities posted during the period from Jan. 1, 2019 to Jul. 1, 2021.
Based on consideration of the timeliness and convenience of related data, the center created about 10,000 datasets based on about 550,000 data samples.
The center categorized hate speech into eight groups, including ‘female/family,’ ‘sexual minorities,’ ‘male,’ ‘race/nationality,’ ‘age,’ ‘region,’ ‘religion,’ and ‘others.’
Smilegate also presented a standard model for the classification of online hate speech. The company intends to unveil the collected data through the GitHub page of its AI center within this month.
Kevin Lee (kevinlee@koreabizwire.com)
