SEOUL, Feb. 22 (Korea Bizwire) – Two major South Korean web portals – Naver and Daum – have introduced artificial intelligence mechanisms that can recommend news articles to viewers.
Article arrangement is a crucial part of the news editing process, and considering that Naver and Daum are two of the most popular platforms for South Koreans to read news articles, many industry watchers are hesitant to accept the latest advancement in robot journalism.
Naver’s AI platform AIRS (AI Recommender System) is the newer of the two, and was just launched last Friday. It is only in its beta stage, available to a group of randomly selected users. AIRS is limited to Naver’s mobile platform at the moment, and is based on two key technologies – collaborative filtering and recurrent neural networks.
Collaborative filtering recommends articles based on specific groups of users who share similar interests. For instance, if “A” reads an article on illegal immigration, similar articles read by “B” and another article by “C” are recommended to “A”, and vice versa.
On the other hand, a recurrent neural network studies a user’s specific reading pattern, such as reading an article on weather after reading about politics, and takes that into consideration when recommending article.
Daum’s AI platform, RUBICS (Real-time User-Behavior Interactive Content Recommender System) is similar to Naver’s recommendation mechanism, allocating news articles on the news feed based on personal preferences like favored writers or articles read by similar age groups sharing specific interests.
RUBICS was introduced in June 2015, and is used on both mobile and PC platforms.
The concerns that pundits have for these mechanisms mainly target two things – political or social bias, which local web portals have long been criticized for but who now have machines to blame it on, and excessive commercialization of articles to serve the companies’ financial interests, not the interests of the general public.
“It may be convenient for consumers who would have easier access to information that they prefer, but at the same time there’s a danger of blocking information that may be more relevant, which can’t be said is knowledge that they don’t want,” said professor Nam Jae-il, from the department of mass communications at Kyungpook National University.
Another mass communications professor, Kim Seo-joong from Sungkonghoe University, said that “the fundamental value of news media, as a means of communication for the society, can be threatened if it only provides prejudiced information.”
Indiscriminate gathering of private information by the AI mechanisms is subject to controversy as well.
“To recommend news articles, they’ll need the server log information,” said senior researcher Kim Wi-geun with the Korea Press Foundation. “Web portals can use this information to visualize people’s interests and political orientation.”
Meanwhile, Korean web portals continue to emphasize their identity as IT companies despite their undeniable role as news media outlets. In a 2015 study, over 80 percent of South Koreans said that they read news articles on internet portals.
By Joseph Shin (email@example.com)