新闻推荐必然导致“茧房”效应吗?--基于模拟新闻平台的实验研究  

Does News Recommendation Inevitably Lead to a Cocoon Effect?--An Experimental Study of a Simulated News Platform

作　　者：刘茜 汤清扬 闵勇 傅晨波[2] 赵芸伟 LIU Qian;TANG Qing-yang;MIN Yong;FU Chen-bo;ZHAO Yun-wei

机构地区：[1]北京师范大学新闻传播学院,北京100088 [2]浙江工业大学网络空间安全研究院,浙江杭州310014 [3]国家计算机网络与信息安全管理中心,北京100029

出　　处：《新闻大学》2023年第2期28-43,118,共17页Journalism Research

摘　　要：新闻推荐算法是否是导致“信息茧房”的罪魁祸首目前还存在争议,尤其是缺乏直接的实验证据和科学的操作化研究。为验证新闻推荐算法的具体特性在用户信息消费多样性中实际扮演的角色,本研究设计和开发了一个能够模拟不同信息分发模式和推荐自由度的实验新闻应用,并开展了一项涉及113名参与者的实验室控制实验研究。实验结果并不支持将“信息茧房”完全归罪于推荐机制本身,而是更为强调用户活动和算法逻辑两个层面的人为因素对“信息茧房”的影响。Whether recommendation algorithms are responsible for the information cocoon is controversial,primarily because of the lack of direct experimental evidence and accurate operationalization.To verify the role that specific characteristics of recommendation algorithms play in the diversity of user information consumption,the researchers developed a news application that simulates different information distribution patterns and degrees of recommendation freedom and conducted a controlled laboratory experiment involving 113 participants.The current experimental results do not support blaming the information cocoon entirely on the recommendation mechanism itself but rather emphasize the human influence at user activity and algorithmic logic levels.

关 键 词：推荐算法 信息茧房 新闻多样性 

分 类 号：G206[文化科学—传播学] TP391[自动化与计算机技术—计算机应用技术]