[Paper Review] Privacy Intelligence: A Survey on Image Sharing on Online Social Networks.

Chi Liu, Tianqing Zhu+2 more|arXiv (Cornell University)|Aug 27, 2020

Privacy, Security, and Data Protection4 citations

TL;DR

This paper proposes a systematic framework for privacy intelligence in online social network (OSN) image sharing, organizing the lifecycle into local, online, and social experience stages. It identifies privacy risks at each stage, reviews intelligent solutions, and outlines a closed-loop privacy-enhancing chain, highlighting challenges and future directions in datasets and system design.

ABSTRACT

Image sharing on online social networks (OSNs) has become an indispensable part of daily social activities, but it has also led to an increased risk of privacy invasion. The recent image leaks from popular OSN services and the abuse of personal photos using advanced algorithms (e.g. DeepFake) have prompted the public to rethink individual privacy needs when sharing images on OSNs. However, OSN image sharing itself is relatively complicated, and systems currently in place to manage privacy in practice are labor-intensive yet fail to provide personalized, accurate and flexible privacy protection. As a result, an more intelligent environment for privacy-friendly OSN image sharing is in demand. To fill the gap, we contribute a systematic survey of 'privacy intelligence' solutions that target modern privacy issues related to OSN image sharing. Specifically, we present a high-level analysis framework based on the entire lifecycle of OSN image sharing to address the various privacy issues and solutions facing this interdisciplinary field. The framework is divided into three main stages: local management, online management and social experience. At each stage, we identify typical sharing-related user behaviors, the privacy issues generated by those behaviors, and review representative intelligent solutions. The resulting analysis describes an intelligent privacy-enhancing chain for closed-loop privacy management. We also discuss the challenges and future directions existing at each stage, as well as in publicly available datasets.

Motivation & Objective

• Address the growing risk of privacy invasion due to image sharing on online social networks (OSNs), especially amid rising incidents of leaks and DeepFake misuse.
• Identify limitations in current OSN privacy systems, which are labor-intensive and lack personalization, accuracy, and flexibility.
• Propose a comprehensive, lifecycle-based framework for intelligent privacy management in OSN image sharing, covering local, online, and social experience stages.
• Systematically analyze privacy issues arising from user behaviors at each stage and evaluate representative intelligent solutions.
• Highlight open challenges and future research directions, including dataset availability and system integration

Proposed method

• Develop a three-stage lifecycle framework: local management (pre-sharing), online management (during sharing), and social experience (post-sharing) for OSN image sharing.
• Map typical user behaviors at each stage—such as image selection, metadata handling, and social feedback—to specific privacy risks like exposure to unauthorized parties or re-identification.
• Review representative intelligent solutions, including AI-driven content analysis, access control policies, and dynamic privacy-preserving transformations.
• Integrate findings into a closed-loop privacy-enhancing chain that enables adaptive, context-aware privacy management across the entire image sharing lifecycle.
• Analyze existing publicly available datasets to assess their suitability for training and evaluating privacy intelligence systems.
• Identify technical and systemic challenges at each stage, such as scalability, user adaptability, and real-time processing needs

Experimental results

Research questions

• RQ1 What are the key privacy risks introduced during each stage of the OSN image sharing lifecycle—local, online, and social experience?
• RQ2 How can intelligent systems effectively address privacy issues arising from user behaviors such as image selection, metadata sharing, and social interaction?
• RQ3 What are the most effective intelligent solutions for achieving personalized, accurate, and flexible privacy protection in OSN image sharing?
• RQ4 How can a closed-loop privacy-enhancing system be designed to adapt across the entire image sharing lifecycle?
• RQ5 What challenges remain in dataset availability, system scalability, and real-world deployment for privacy intelligence in OSN image sharing?

Key findings

• Current OSN privacy systems are insufficient due to their labor-intensive nature and lack of personalization, accuracy, and flexibility in protecting user privacy.
• The proposed three-stage lifecycle framework effectively structures privacy challenges and solutions across local, online, and social experience phases.
• Intelligent solutions such as AI-based content analysis and dynamic access control show promise in mitigating privacy risks at scale.
• A closed-loop privacy-enhancing chain enables adaptive, context-aware protection that evolves with user behavior and social feedback.
• Significant challenges remain in dataset availability, system integration, and real-time performance for deploying privacy intelligence solutions.
• Future research must prioritize interoperable, user-centric systems that balance privacy, usability, and functionality in OSN image sharing