Bflysoft is showing strong performance. It appears that the company is gaining attention for possessing the optimal artificial intelligence (AI) training data, namely 'news data.'
Bflysoft is a company that provides media platform services based on news copyrights and AI technology. It has signed a copyright distribution contract with the Korea Press Foundation, a copyright trust organization for media companies. The company offers news copyright distribution, media big data analysis, personalized news services, and AI training data. Based on systematic distribution and legal rights acquisition, it has collected and supplied content from over 3,000 domestic media outlets for about 20 years.
They introduced 'RDP Line,' which provides vast news data accumulated through the 'iRoot' technology that digitizes print news and the iSurfer platform for algorithm training. Bflysoft stated that it has established real-time data labeling technology using AI 24/7 through the largest media alliance in Korea. They also mentioned that by refining and processing various media such as newspapers, broadcasts, online news, and magazines, they possess the largest volume of AI training data in the country. Verified data services such as Naver News Library, Korea Newspaper Archive, and real-time API services are provided.
Bflysoft holds approximately 700 million accumulated articles as well as 200,000 articles generated daily. The documents are composed of around 2,000 characters, suitable for LLM training, and cover various categories such as politics, economy, culture, and arts, making them valuable as domain-specific tuning data. Unlike documents, papers, and books that compress core information, these short 2,000-character articles contain essential content. The ability to update data in real time is also cited as a unique strength of Bflysoft. Regarding data reliability, Bflysoft explained that there are very few data sources that surpass news.
Geonjae Lee, a researcher at IBK Investment & Securities, explained, "If the AI training data market fully blooms, there is a possibility that companies supplying high-quality news data will stand out."
As AI develops, the value of data is increasing. ChatGPT learns from news by the U.S. news agency AP News. OpenAI, the developer of ChatGPT, and AP News have signed a license agreement regarding the use of news articles. The license agreement is interpreted as a response to copyright controversies surrounding the vast amount of information, including news and books, necessary for training generative AI.
Marc Benioff, CEO of Salesforce, recently stated in an interview during the Swiss Davos Forum, "AI companies have stolen intellectual property (IP) to build technology," and pointed out, "AI companies need to standardize payment methods to fairly treat content creators."
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
