寻找新朋友的地点和时间:基于位置的社交发现网络分析The Where and When of Finding New Friends: Analysis of a Location-Based Social Discovery Network |
|
课程网址: | https://videolectures.net/videos/icwsm2013_chen_discovery_network |
主讲教师: | Terence Chen |
开课单位: | 信息不详。欢迎您在右侧留言补充。 |
开课时间: | 2014-04-03 |
课程语种: | 英语 |
中文简介: | 随着越来越多的人使用移动设备访问在线社交网络,基于位置的功能已经成为社交网络的重要组成部分。在本文中,我们对一个新的基于位置的在线社交网络服务类别进行了第一次测量研究,这是一个基于位置的社交发现网络(location-based social discovery,简称LBSD),它让用户能够发现附近的人并与他们交流。与流行的基于签到的社交网络不同,LBSD允许用户公开自己的位置,而不必与特定的“场所”关联,而且他们的使用不受潜在虚拟社区激励机制的影响。通过分析从一个流行的新LBSD网络收集的800多万个用户资料和大约1.5亿个位置更新,我们首先展示了观察用户的时空使用模式特征,显示40%的更新来自用户的主要位置,80%来自用户的前10个位置。我们发现了引发用户数量激增的事件,显示了社交媒体营销的重要性。最后,我们研究了如何利用使用模式来重新识别具有不同标识符的个人,或从属于不同在线服务的数据集中识别的个人。我们通过使用情况、空间和时空模式以及使用大量指标来评估重新识别的效果,结果表明,使用位置数据可以获得高精度的最佳结果:我们的实验表明,我们可以使用监控的空间数据以77%的精度重新识别多达85%的用户。总体而言,我们发现尽管用户在使用模式和行为上表现出强烈的周期性行为,但重新识别的成功率在很大程度上取决于活跃程度和用户在网络中的寿命。 |
课程简介: | With more people accessing Online Social Networks (OSN) using their mobile devices, location-based features have become an important part of the social networking. In this paper, we present the first measurement study of a new category of location-based online social networking services, a location-based social discovery (LBSD) network, that enables users to discover and communicate with nearby people. Unlike popular check-in-based social networks, LBSD allows users to publicly reveal their locations without being associ- ated to a specific “venue” and their usage is not influenced by the incentive mechanisms of the underlying virtual community. By analyzing over 8 million user profiles and around 150 million location updates collected from a popular new LBSD network, we first present the characteristics of spatial- temporal usage patterns of the observed users, showing that 40% of updates are from the user’s primary location and 80% are from their top 10 locations. We identify events that trigger bursts of growth in subscriber numbers, showing the importance of social media marketing. Finally, we investigate how usage patterns may be utilized to re-identify individuals with e.g. different identifiers or from datasets belonging to different online services. We evaluate re-identification by usage, spatial and spatial-temporal patterns and using a number of metrics and show that the best results can be achieved using location data, with a high accuracy: our experiments demonstrate that we can re-identify up-to 85% of users with a precision of 77% using monitored spatial data. Overall, we find that although users exhibit strong periodic behavior in their usage pattern and movements, the success rate of re-identification is highly dependent on the level of activeness and the lifetime of the users in the network. |
关 键 词: | 社交网络; 社交媒体营销; 使用模式 |
课程来源: | videolectures |
数据采集: | 2025-03-23:yuhongrui |
最后编审: | 2025-03-23:yuhongrui |
阅读次数: | 5 |