Bflysoft Completes Construction of AI-Based 'Clickbait Article Detection Data'

[Asia Economy Reporter Minji Lee] Bflysoft announced on the 17th that it has successfully completed the "2022 AI Training Data Construction Project for Clickbait Article Detection," led by the Ministry of Science and ICT and promoted by the National Information Society Agency (NIA).

This project was conducted to address the social issues caused by clickbait articles and to secure high-quality AI training data aimed at restoring trust in journalism.

Bflysoft categorized the data into articles with mismatched titles and body content, articles lacking domain consistency in the body, and non-clickbait articles for training purposes, building a total of 720,000 data entries and developing two AI language models capable of detecting such articles. These models are scheduled to be released on the NIA's AI Hub in the first half of 2023.

A Bflysoft representative stated, "Through this project, we have completed the development of the largest-scale article classification and detection data in Korea, along with top-tier AI language models. We expect these to be usefully applied by portal and media platform operators, as well as media-related regulatory bodies, contributing to the advancement of the media industry."

Bflysoft has continuously enhanced AI language models specialized for media, such as spell checking, automatic article classification, and article summarization, integrated into its proprietary ‘WIGO AI platform,’ expanding supply to its own services and media networks to improve satisfaction with media services.

Meanwhile, Bflysoft, which provides media-specialized service platforms such as iSurfer and Roseus, announced plans to focus its capabilities on expanding into a global media subscription platform based on 20 years of accumulated media big data and language AI technologies including natural language processing.