Game publisher Smilegate establishes datasets for hateful online comments

By Park Sae-jin Posted : January 20, 2022, 16:52 Updated : January 20, 2022, 16:52

[Courtesy of Smilegate]

SEOUL -- Smilegate, a South Korean game publisher, has created a dataset that contains hateful online comments and speeches. The dataset can be used to prevent cyberbullying and online discrimination.
Cyberbullying has long been a major problem in South Korea's internet world. Popular female celebrities Sulli and Goo Hara have committed suicide after being attacked by cyberbullies including radical feminists in 2019. At least one percent of students have suffered from online violence, according to South Korea's education ministry.
Smilegate said in a statement on January 20 that it would work with Underscore, a  knowledge content startup, to release the dataset through GitHub, a global code hosting platform, by the end of January. The dataset can be used in many online industries including internet-based gams, chatbot services, and online surveys.
The game publisher said that some 550,000 data samples have been collected from portal websites and online community posts for about 30 months from 2019 to create about 10,000 datasets. Smilegate said it would continue to upgrade its data-collecting capability to secure a technique that can detect hateful online activities more accurately.
기사 이미지 확대 보기