应对过滤气泡:算法策展对用户信息消费行为选择性和态度极端化的影响OA北大核心CHSSCDCSSCICSTPCD

[Purpose/Significance]There is a growing concern that algorithms have led to"filter bubbles"in which people only receive content tailored to their existing inclinations,as opposed to being exposed to diverse information.And the filter bubble is reflected in the selectivity of information consumption and attitude extremity.This study aims to explore the intervention effects of different approaches of prioritization in recommender systems on filter bubbles.[Method/Process]This study recruited 38 participants to visit the mock personalized news recommender system.And they were assigned to groups with different prioritization approaches.This study observed and recorded each participant's clicking behavior and reading behavior through server logs.Additionally,the paper measured the changes in participants'attitude extremity using a scale.[Result/Conclusion]The results show that the approaches of prioritization did not affect the selectivity of users'clicking behavior,but affect the selectivity of users'browsing behavior and change of attitude extremity.Specifically,both time-based ranking and quality-based ranking significantly reduce the selectivity of users'reading behaviors when compared to preference-based ranking.And quality-based ranking also weakens the extremity of users'attitudes to a greater extent.This study not only provides a new research perspective and effective methodology for filter bubble-related research,but al-so provides practical insights for recommendation systems algorithm design.