RePriCo: Resolving Multi-party Privacy Conflicts in Social Media

上传到社交媒体的数千亿个项目由多个用户共同拥有，但只有上传项目的用户才能设置其隐私设置(即，谁可以访问该项目)。这是一个巨大而严重的问题，因为用户对共有物品的隐私偏好通常会发生冲突，因此只应用一方的偏好会导致侵犯隐私的行为，并带来严重的后果(例如，用户失去工作，被网络跟踪等)。对这一问题的解决是最及时的，因为它已被强调为适当的在线隐私管理需要解决的最重要问题之一。这个问题的一个例子是一张照片，其中Alice和Bob被描绘在一起：他们如何就他们将与之共享照片的第三方达成一致？主流社交媒体只允许Alice(假设她上传了这篇文章)设置照片的隐私设置，但如果Bob不想与Alice的一些朋友分享怎么办？在过去的几年里，已经提出了一些初步的方法，但这些方法有一些限制，使它们在实践中不适用。用户被迫尝试手动解决此类冲突(通过电子邮件、电话等)，这令人精疲力竭，因为社交网络非常动态且规模巨大。更重要的是，手动解决冲突的过程开始得太晚了，此时一个用户已经发布了该项目，并且发生了侵犯隐私的行为。RePriCo最令人兴奋、最新颖和最具开创性的部分将是基于新的自动协商技术的自动冲突检测和解决机制。特别是，Reprico的主要挑战将是以这样一种方式自动建议冲突的解决方案，即这些解决方案在大多数时间被用户接受，并且他们不需要手动解决冲突。这既是雄心勃勃的，也是冒险的，因为它需要了解如果用户自己手动解决冲突，他们实际上将如何达成协议。Reprico的假设是，用户的协商行为受到用户关系及其强度的显著影响(现有证据似乎支持这一点)，计算协商机制应该基于此来产生可接受的结果。Reprico的主要贡献之一将是关于用户关系如何影响谈判行为的强有力的经验基础。Reprico将在许多方面具有极大的变革性：(I)它将使用户能够管理多方共同拥有物品的隐私；(Ii)它将考虑用户实际可以接受的内容以及导致这一点的因素，这是以前解决多方隐私冲突的方法从未考虑过的；以及(Iii)它将建立的经验基础不仅将影响Reprico的自动冲突检测和解决机制，还将有助于推动对该主题的研究。

Hundreds of billions of items that are uploaded to Social Media are co-owned by multiple users, yet only the user that uploads the item is allowed to set its privacy settings (i.e., who can access the item). This is a massive and serious problem as users' privacy preferences for co-owned items usually conflict, so applying the preferences of only one party leads to privacy violations with severe consequences (e.g., users losing their jobs, being cyberstalked, etc.). A solution to this problem is most timely as it has been highlighted as one of the most important problems to be addressed for an appropriate online privacy management. An example of this problem is a photo in which Alice and Bob are depicted together: how can they agree on the third parties they will share the photo with? Mainstream Social Media would only allow Alice (assuming she uploads the item) to set the privacy settings for the photo, but what if Bob would not like to share with some of Alice's friends? Some initial approaches have been proposed in the past few years, but these have a number of limitations that make them unsuitable in practice. Users are forced to try to resolve such conflicts manually (by e-mail, phone calls, etc.), which is exhausting because social networks are very dynamic and huge in scale. Even more importantly, the process of resolving conflicts manually starts too late, when one user has already posted the item and the privacy violation has occurred.RePriCo's most exciting, novel, and ground-breaking part will be the automated conflict detection and resolution mechanism based on new automated negotiation technologies. In particular, RePriCo's main challenge will be to automatically suggest solutions to the conflicts in such a way that these solutions are accepted by users most of the time and they do not need to resolve the conflicts manually. This is both ambitious and adventurous, as it requires understanding how users would actually reach an agreement if they were to solve the conflicts themselves manually. RePriCo's hypothesis is that users' negotiation behaviour is significantly influenced by users' relationships and their strength (as existing evidences seem to support), and that a computational negotiation mechanism should be based on this in order to produce acceptable outcomes. One of the main contributions of RePriCo will be a strong empirical base about how users' relationships influence negotiation behaviour. RePriCo will be highly transformative in many aspects: (i) it will empower users to manage their privacy for multi-party co-owned items; (ii) it will consider what is actually acceptable for users and the factors that contribute to that, which no previous approach to solve multi-party privacy conflicts has considered before; and (iii), the empirical base it will develop will not only influence RePriCo's automated conflict detection and resolution mechanisms but also help push research on the topic forward.