News and user characteristics used by personalized algorithms: The case of Korea's news aggregators, Naver News and Kakao News

Abstract

Digital intermediaries, such as search engines and news aggregators, have recently implemented personalized algorithms as a method for recommending news articles to their users. This study is focused on Korea's representative news aggregators, Naver News and Kakao News, to discern differences between news characteristics when personalized algorithms are applied and when they are not, and analyze which user characteristics have an effect on the news articles recommended by personalized algorithms. The results are as follows. There was a difference in the topics of news articles selected when personalized algorithms were applied and when they were not, with articles selected by algorithms displaying a greater diversity of news sources. News articles selected by personalized algorithms included fewer articles with sizeable audience feedback, press endorsements, and recency. When selecting news articles, personalized algorithms were shown to be affected not only by the demographic characteristics of users but also by their political orientation and level of news consumption. In particular, users who consumed less news or had politically conservative tendencies were provided with news articles that were differentiated from the news articles of users who did not use personalized algorithms.