应对行为视角下算法推荐内容相似性对App持续使用的影响  

The Effect of Recommendation Content Similarity on Sustained App Use from the Perspective of User Coping

作　　者：廖觅燕 方佳明[1] 杨晶晶 Altab Hossin Liao Miyan;Fang Jiaming;Yang Jingjing;Altab Hossin(University of Electronic Science and Technology;Chengdu University)

机构地区：[1]电子科技大学经济与管理学院 [2]成都大学创新创业学院

出　　处：《南开管理评论》2023年第3期178-188,I0032,I0033,共13页Nankai Business Review

基　　金：国家自然科学基金项目(71571029)资助。

摘　　要：App推荐算法在实现用户和推荐内容精准匹配的同时也会导致“信息茧房”效应,加剧用户疲惫体验。基于用户应对过程理论,本研究将App用户的失实交互行为视为用户应对推送内容疲惫的一种重要“技术适应”,构建了移动App场景下算法推荐内容相似性用户应对模型。通过访谈和多轮问卷调查分析发现,算法推送内容相似性导致的疲惫体验与知觉控制感共同决定了App用户随后采取的应对努力策略。问题聚焦应对正向促进了失实交互行为,而情绪聚焦应对则对失实交互行为具有负面影响。失实交互行为中介了问题聚焦应对与情绪聚焦应对行为对用户App持续使用的影响。研究结果打开了用户疲惫体验与产品持续使用之间的理论黑箱,丰富和拓展了现有在线用户疲劳感研究和用户适应应对理论。研究结果对移动端推荐系统的开发与改进具有实践参考价值。In the age of mobile connectivity,the popularity of mobile smart devices,including smartphones and tablets,is skyrocketing.They have become the primary means for most users to access the Internet.To stay competitive in the market,many mobile application(app)companies have embraced an algorithm-driven interactive content distribution model.This model aims to push personalized content to enhance user experience and increase user engagement.Currently,algorithm-based personalized content distribution accounts for 70%of the entire Internet’s information flow.A prime example of this approach is seen in TikTok’s algorithmic recommendation system.This system captures and learns from almost every user action,such as completion rates,likes,follows,comments,and favorited content.The feedback gathered is then used to improve the accuracy of the recommendation results.In a personalized recommendation system,accurately capturing user preferences is a critical process that directly impacts the quality of subsequent interest matching and recommendation list sorting.To achieve this,algorithms collect user usage data,such as likes,comments,and shares,as inputs to the content push model.This enables accurate matching of user preferences with relevant content.By leveraging these algorithms,companies can tailor their content delivery to each user,creating a more engaging and personalized experience.The content distribution model driven by app recommendation algorithms aims to achieve accurate matching of users and content.However,it also has an impact on users’independent choice of information.The algorithm tends to recommend content that aligns with users’existing preferences,resulting in a more centralized distribution of information.This phenomenon can potentially lead to the“information cocoon”effect and contribute to user fatigue.However,the literature so far has not been able to explain why some users continue to use the product even when they feel fatigued.In fact,in order to cope with the fatigue caused by the sim

关 键 词：移动App 内容推荐算法 信息茧房 失实交互 AI算法欺骗 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术] TP311.56[自动化与计算机技术—计算机科学与技术]