[4H1-OS-2a] ニュースメディアのデータサイエンス(1/2)

[4H1-OS-2a-03] Effect of Article Diversity on retention rates in Online News Service

〇Shusuke Suganuma¹, Kojiro Iizuka², Yoshifumi Seki², Fujio Toriumi¹ (1. The University of Tokyo, 2. Gunosy Inc.)

Keywords:news, recommender system, user engagement

While the problems of filter bubbles and echo chambers caused by recommendation systems on digital platforms and the resulting risk of confusion are well understood, the impact of diverse information recommendations on user engagement has not been well analyzed. In this study, we analyzed user behavior logs of a popular news application to determine how the degree of diversity of the news provided by the application affects the user's continuous usage rate. As a result, we found that users who read homogeneous articles had a significantly lower retention rate. At the same time, it was suggested that users who prefer extremely homogeneous articles have a different behavioral pattern.