print PRINT

SCIENCE > Technology

AI able to discern online posts by minors

  • May 24, 2017
  • , Yomiuri evening edition , p. 10
  • JMH Translation

A research team of the University of Tokyo and IT firm CyberAgent developed a technology that uses AI to analyze online posts and detect those made by minors. It is attracting attention for its potential to protect adolescents from cybercrime and internet predators. The technology was presented in an AI conference held in Nagoya on May 24.


CyberAgent used 10 million posts made to a social networking site run by its subsidiary, and had its AI learn the content of the postings and the ages of the contributors. As a result, AI was able to predict with 90% accuracy whether the post was written by a minor. AI analyzes the vocabulary used in the messages. Minors, for example, tend to use words associated with school life and extracurricular activities such as ‘homework.’ AI categorizes the posts by age group.


“We hope to use this to protect minors in social networking sites,” says University of Tokyo Associate Professor Fujio Toriumi of the School of Engineering.


AI can only analyze public postings, so the detection feature cannot be used on private platforms such as LINE, a free instant communication app.


“While we must ensure the privacy of communications, we have now arrived at a time when we need to reevaluate how socially beneficial information is managed,” points out Professor Masatomo Suzuki of Niigata University who specializes in information law.

  • Ambassador
  • Ukraine
  • COVID-19
  • Trending Japan