Privacy Intelligence: A Survey on Image Privacy in Online Social Networks

View on arXiv ← Back to list

Authors: Chi Liu, Tianqing Zhu, Jun Zhang, Wanlei Zhou

Published: 2020-08-27 15:52:16+00:00

AI Summary

This paper surveys privacy intelligence solutions for online social network (OSN) image sharing. It presents a taxonomy of OSN image privacy and a lifecycle-based framework for analyzing privacy issues and intelligent solutions, culminating in an "intelligent privacy firewall."

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 in OSN image sharing. However, OSN image privacy itself is quite complicated, and solutions currently in place for privacy management in reality are insufficient to provide personalized, accurate and flexible privacy protection. A more intelligent environment for privacy-friendly OSN image sharing is in demand. To fill the gap, we contribute a survey of privacy intelligence that targets modern privacy issues in dynamic OSN image sharing from a user-centric perspective. Specifically, we present a definition and a taxonomy of OSN image privacy, and a high-level privacy analysis framework based on the lifecycle of OSN image sharing. The framework consists of three stages with different principles of privacy by design. At each stage, we identify typical user behaviors in OSN image sharing and the privacy issues associated with these behaviors. Then a systematic review on the representative intelligent solutions targeting those privacy issues is conducted, also in a stage-based manner. The resulting analysis describes an intelligent privacy firewall for closed-loop privacy management. We also discuss the challenges and future directions in this area.

Key findings

The survey categorizes OSN image privacy into observable, inferential, and contextual privacy. It identifies common design principles for privacy intelligence at each stage of the image lifecycle: offline mode, sensitivity, and preventive intelligence (local); online mode, visibility, and protective intelligence (online); in-the-wild mode, contextual integrity, and persistent intelligence (social). The authors highlight challenges like the privacy-utility trade-off and the need for more robust solutions against DeepFakes.

Approach

The authors propose a lifecycle-based privacy analysis framework encompassing local management, online management, and social experience stages. For each stage, they identify privacy issues arising from typical user behaviors and review intelligent solutions, focusing on automated and AI-driven approaches.

Datasets

PicAlert, YourAlert, VISPR, VISPR-extension, VizWiz-Priv

Model(s)

Convolutional Neural Networks (CNNs), Graph Neural Networks (GNNs), Generative Adversarial Networks (GANs), Ensemble learning, various machine learning classifiers

Author countries

Australia, China

← Previous