EAGER: Towards a Better Understanding of Group Privacy in Social Media Community Detection

如今，人类的大部分交流都是通过在线社交网络进行的。Twitter、Facebook和Youtube现在争夺我们的集体注意力，就像电视、广播和报纸争夺前几代人的注意力一样。但当代在线社交媒体与过去的媒体有质的不同。在线交流留下了谁对谁说了什么，什么时候说的，关于什么话题的记录。新的分析工具的开发提供了使用这些记录来跟踪流行的在线主题并确定对这些主题作出贡献的群体的人口统计数据的可能性，包括贡献者的地理位置以及他们的年龄，性别和种族。此外，特设小组有可能在真实的时间内围绕主题进行联合。一方面，这些数据对计算机科学家提出了挑战，要求他们开发新的工具来跟踪这些类型的信息。在这一领域的成功在商业、市场营销和政治上都有着重要的实际利益。然而，与此同时，跟踪这类信息的能力引起了隐私问题，无论是对个人还是对可以通过新兴技术识别的群体成员。在我们的研究中，计算机科学家将开发能够跟踪主题和组成员的工具，通信研究人员将确定人们围绕这些信息产生的隐私问题的类型，何时，为什么以及后果。该项目中描述的研究建立在趋势分析和社区提取的先前工作的基础上，寻求在两个方面推进研究：有效地识别专注于流行话题的特设社区，以及基于所识别的社区来理解个人与群体隐私。尽管趋势分析是计算机科学中的新兴研究领域，但现有模型仅关注一维的相关性（例如，位置）与趋势主题。这项研究代表了一个进步，提供了一个更复杂和强大的工具，允许提取有趣的和潜在有用的趋势模式，通过基于主题的社区识别。与此同时，这种社区认同可能会引发社会科学中尚未研究的新型群体隐私问题，社会科学主要关注个人隐私而不是群体隐私。这种方法提供了一个独特的机会，显着影响学术理解的机制和动态的个人和群体的隐私问题，特别是关于特设的，基于主题的群体，以及他们对社交媒体用户的态度和行为的影响。如果组级别的隐私是一个超越个人隐私的关注，那么我们希望发现，人们会表达群体隐私的关注时，他们认同的一组是包括在跟踪信息中，特别是当主题是道德加载，但独立的个人参与者是否亲自参与的Twitter对话。这有可能为群体隐私问题的出现创造必要和充分的条件。

Much of human communication is now mediated by online social networks. Twitter, Facebook, and Youtube now compete for our collective attention in much the same way as television, radio, and newspapers did for previous generations. But contemporary online social media are qualitatively different from media of the past. Online communication leaves a record of who said what to whom, when, and on what topic. The development of new analytical tools offer the possibility to use these records to track popular on-line topics and to identify the demographics of groups contributing to these topics, including the geographic location of contributors, as well as their age, gender, and ethnicity. What is more, it is possible for ad hoc groups to coalesce around topics in real time. On the one hand, these data present a challenge for computer scientists to develop new tools that enable the tracking of these kinds of information. Success in this domain offers significant practical benefits in business, marketing, and politics. At the same time, however, the ability to track these kinds of information raise privacy concerns, both for individuals and for members of groups who can be identified by the emerging technology. In our research, computer scientists will develop tools that enable tracking of topics and group memberships, and communication researchers will identify the kinds of privacy concerns that people develop around these kinds of information, when, why, and with what consequences.The research described in this project builds on prior work on trend analysis and community extraction, seeking to advance research on two fronts: the efficient identification of ad hoc communities which focus on a popular topic, and the understanding of individual versus group privacy based on the identified communities. Although trend analysis is a burgeoning area of research in Computer Science, existing models focus on the correlation of only one dimension (e.g., location) with trending topics. This research represents a step forward by providing a more sophisticated and powerful tool that allows for the extraction of interesting and potentially useful trend patterns through Topic Based Community Identification. At the same time, this community identification may prompt new types of group privacy concerns that have not been researched in social science, which has mainly focused on individual rather than group privacy. This approach provides a unique opportunity to significantly impact scholarly understanding of the mechanisms and dynamics of individual and group privacy concern, especially with regard to ad-hoc, topic-based groups, and their effects on Social Media users' attitudes and behavior. If group-level privacy is a concern beyond individual privacy, then we expect to find that people will express group privacy concerns when a group they identify with is included in the tracking information, especially when topics are morally loaded, but independent of whether the individual participant is personally involved in the Twitter conversation. This has the potential to develop necessary and sufficient conditions for the emergence of group privacy concerns.