Problem Summary

In the realm of online social networks, photo sharing involves multiple stakeholders whose privacy needs to be collaboratively managed. Collaborative privacy management in photo sharing requires fine-grained control, context-aware enforcement, and scenario-based policies to protect user privacy effectively.

Rationale

Existing privacy controls are too coarse-grained and do not adequately protect individual elements within shared photos. Users need more intuitive and automated systems to manage their privacy effectively without significant effort.

Solution

Enhance privacy protection in social networks by providing more detailed, context-aware, and collaborative privacy controls for photo sharing.

Shu, Zheng and Hui [1] proposed Cardea, a context-aware visual privacy protection system designed for photos taken and shared via mobile and wearable devices. It protects visual privacy based on user-specified preferences related to location, scene, presence of others, and hand gestures. Cardea can be integrated into camera apps and social media platforms to enforce privacy measures such as blurring faces automatically.

Li et al. [2] presented HideMe, a framework for privacy-preserving photo sharing on social networks. It allows users to set scenario-based privacy policies, automatically blurring faces based on user-defined conditions such as time, location, and relationships. HideMe includes a distance-based algorithm to protect bystanders' privacy and an efficient face-matching algorithm to reduce system overhead. HideMe prototype is available at GitHub: https://github.com/HideMe2018/HideMe.

Vishwamitra et al. [3] proposed a PII-based Multiparty Access Control (PMAC) model to address the privacy concerns in photo sharing on Online Social Networks (OSNs). This model enables fine-grained control over Personally Identifiable Information (PII) within shared photos. The PMAC model includes a policy specification scheme and a policy enforcement mechanism, allowing multiple users to manage access to their PII items collaboratively.

Lin et al. [4] presented a mechanism called REMIND to estimate the risk of privacy breaches when sharing images on social networks. REMIND uses a probabilistic model to evaluate the likelihood of unwanted image disclosure based on various factors, including user behaviour and image content. If the computed probability indicates a high risk of privacy breach, the image owner is reminded to help revise privacy settings and harmonise policies for multi-owner images​.

Platforms: personal computers, mobile devices

Related guidelines: Implement Collaborative Privacy Management for Shared Data in Social Networks

Example

Cardea privacy preference setting interface and Privacy protection example [1]. (See enlarged)

• The integrated system demonstrates an 86% overall accuracy in protecting privacy, indicating promising potential for context-aware visual privacy protection [1]. Evaluations show that solutions like HideMe [2] effectively maintain privacy while ensuring system efficiency. A prototype implementation on Facebook underscores the feasibility and practicality of enhancing user privacy controls [3]. Additionally, models such as REMIND [4] can be seamlessly applied to various types of co-owned or co-managed content in online social networks, extending beyond just images.

• Integrating mechanisms like Cardea into existing camera apps and social media platforms requires significant development effort and cooperation from platform providers [1].