面向关联的推荐系统研究概览
摘要
推荐系统旨在通过基于用户偏好从通用项目集合中选择子集来减少信息超载并保留客户。虽然推荐系统研究源于信息检索和过滤，但该主题已逐渐发展为独立且具挑战性的研究领域。传统上，推荐系统是从基于内容的过滤与协同过滤的设计视角进行研究的。然而，推荐并非在真空中进行，而是置于用户和社交背景的非正式社区中。因此，最终所有推荐系统都在人与人之间建立联系，应从这种视角进行研究。本文因此提出一种以关联为导向的推荐系统研究视角。我们认为推荐本质上具有社交属性，最终旨在通过显式用户建模或通过发现数据中隐含的关系来连接用户。因此，推荐系统的特征在于如何建模用户以将用户联系起来：是显式还是隐式。最终，用户建模和关联导向视角引出广泛和社交问题——如评估、定位以及隐私和信任问题——我们也简要讨论了这些问题。
引用
@article{arxiv.cs/0205059,
title = {A Connection-Centric Survey of Recommender Systems Research},
author = {Saverio Perugini and Marcos Andre Goncalves and Edward A. Fox},
journal= {arXiv preprint arXiv:cs/0205059},
year = {2007}
}
备注
Based on the comments from reviewers, we have made modifications to our article, including the following: Shifted the focus of the survey completely to recommender system research rather than recommendation and personalization and subsequently changed the title to "A Connection-Centric Survey of Recommender Systems Research." Now only cite the most seminal works in this area and as a result have reduced the references significantly from over 200 to 120