用户视角的短视频内容分发平台算法推荐智能化评价  

Intelligent Evaluation Model for Content Platform Recommender System from User Perspective:Scale Development and Multi-case Comparison

作　　者：刘超[1,2] 熊开容 黄欣媛[1] 甘子美 何贻花 LIU Chao;XIONG Kairong;HUANG Xinyuan;GAN Zimei;HE Yihua

机构地区：[1]广东外语外贸大学新闻与传播学院 [2]广州市人文社科重点研究基地“广州城市舆情治理与国际形象传播研究中心” [3]广东工业大学环境科学与工程学院

出　　处：《传媒经济与管理研究》2022年第2期207-234,共28页Media Economics and Management Research

基　　金：教育部人文社科研究项目“湾区形象刻板印象的形成机制与干预策略研究”(19YJA860010);广东省哲学社会科学规划一般项目“关系主义视角的粤港澳大湾区城市形象建构与传播策略研究”(GD19CXW02);广东省普通高校省级重大科研项目“气候谈判背景下中国负责任国家形象对外传播战略研究”(2016WZDXM025);广州城市舆情治理与国际形象传播研究中心课题“粤港澳大湾区城市群的区域形象构建与国际传播研究”(2021-WT-02)。

摘　　要：以抖音、快手、美拍、微视、西瓜视频5个短视频内容分发平台为研究对象,应用关键事件法、德尔菲法及结构方程模型分析发现,准确性、多样性、新颖性、惊喜性是用户视角下反映短视频平台算法推荐智能化效力的四大表征。从算法技术赋能用户使用体验感知的角度而言,目前,各短视频平台在多样性上呈现出最佳的算法技术赋能效果,惊喜性是各平台凭借算法优势收获用户认同的现实短板。更为垂直化的短视频平台借助算法优势,更容易在用户使用体验感知中取得竞争优势。算法技术素养高的用户对短视频平台算法推荐智能化的感知要求更加严苛,男性与低龄人群在四个维度上的评价均显著低于女性和高龄人群。低消费人群对准确性、新颖性和惊喜性的评价显著低于高消费人群,说明用户的消费水平与短视频平台算法推荐智能化感知之间具有群体差异和正向的影响逻辑。算法推荐智能化评价的人口统计特征差异,反映出不同类型短视频平台经由算法赋能后的用户体验感知群像特征。Taking five short video content distribution platforms of Tik Tok,Kuaishou,Meipai,Wesee and Watermelon as the research objects,the key event method,Delphi method and structural equation model were used to analyze the results.It is found that accuracy,diversity,novelty and surprise are the four major characteristics which reflect the intelligent effectiveness of short video platform algorithm recommendation from the user's perspective.From the perspective of algorithm technology empowering users'experience perception,at present,the short video platforms show the best algorithm technology empowerment effect in diversity,and surprise is the reality of the weak link of each platform by virtue of algorithm advantages to gain user's recognition.With the help of algorithm advantages,more vertical short video platforms are more likely to gain competitive advantages in experience perception of users.Users with high algorithm technical literacy have more stringent perception requirements for the intelligence of short video platform algorithm recommendation.The evaluation of males and younger people in the four dimensions is significantly lower than that of females and elderly people.The evaluation of accuracy,novelty and surprise of low consumer population is significantly lower than that of high consumer population,indicating that there is a group difference and positive influence logic between the consumption level of users and the intelligent perception of short video platform algorithm recommendation.The differences in demographic characteristics of the algorithm recommendation intelligent evaluation reflect the user's experience of different types of short video platforms empowered by the algorithm to perceive the characteristics of group images.

关 键 词：算法推荐智能化 个性化推荐 基于用户的算法推荐 用户生产内容 短视频APP 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]