| 罗婷予,Miguel Baptista Nunes.从用户视角理解智能推荐系统[J].数字图书馆论坛,2019,(10):30~36 |
| 从用户视角理解智能推荐系统 |
| Understanding Recommender Systems from Users’ Perspective |
| 投稿时间：2019-10-08 |
| DOI：10.3772/j.issn.1673-2286.2019.10.005 |
| 中文关键词: 推荐系统；用户感知；质性研究；感知因素 |
| 英文关键词: Recommendation Systems; User Perception; Qualitative Study; Impact Factor Study |
| 基金项目:本研究得到广东省自然科学基金面上项目“基于人工智能的虚拟现实古籍理论与模型研究”（编号：2019A1515011260）以及中山大学信息科
学、信息管理和信息系统学科交叉的培育研究科研启动经费（经费号：20000-18841200）资助。 |
| 作者 | 单位 | | 罗婷予 | 中山大学 | | Miguel Baptista Nunes | 中山大学 |
|
| 摘要点击次数: 2647 |
| 全文下载次数: 1692 |
| 中文摘要: |
| 随着推荐系统的广泛应用，探讨用户对推荐系统的理解，了解影响用户感知推荐系统的因素可以为推荐系统的评价以及开发设计提供理论参考。本研究为探索性研究，旨在验证研究的可行性以及研究问题的价值，本文采用定性研究方法，对4名普通用户进行焦点小组访谈，并借助NVivo 11软件通过内容分析法对访谈数据进行分析。研究发现用户感知到的推荐依据有用户偏好、用户行为以及业务需求。感知交互成本、多样性、覆盖率、感知有用性、环境背景兼容性、控制度、灵敏度、跨平台数据共享、推荐精度、新颖性和透明度是影响用户感知推荐系统的因素。 |
| 英文摘要: |
| In order to solve the problems of information overload and information anxiety, recommendation system have been applied widely in many different areas. In order to effectively solve these problems, it is necessary to understand how users perceive recommendation systems and what factors affect their perceptions of usefulness and effectiveness of recommendations. This is very necessary in order to design and evaluate systems. This research is an inductive exploratory pilot study that aims to give a first indication towards understanding what type of factors influence these perceptions. Data was collected through one focus group interview and was then coded using content analysis supported by NVivo 11. The findings of the study reveal that users’ preference, users’ behaviors and business characteristics are 3 criteria which interviewees perceived as important evidence of recommendation. Moreover, interviewees expressed that the factors which affects their perceptions of usefulness and effectiveness of recommendations are perceived interaction effort, diversity of recommendation results, coverage, perceived usefulness, contextual compatibility, perceived controllability, flexibility, cross-platform data sharing, accuracy, novelty, and transparency. This study is not generalizable due to the small numbers of respondents, but provides an initial informed emergent theory that can help designers and developers of recommendation systems improve their efforts. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
