Circle of Trust Construction for Social Network Privacy Leakage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current Online Social Networks (OSNs) lack effective mechanisms to prevent information leakage to unwanted targets, particularly through the 'mention' propagation channel, despite user-selected privacy controls, which can lead to unintended sharing of personal information.
Innovation Solution
A system for constructing a 'circle of trust' on-the-fly, using additive or subtractive methods, including sampling and hybrid approaches, to control information propagation and minimize the probability of information reaching unwanted targets, by estimating potential leakage and optimizing the circle of trust at the time of message posting.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current OSN privacy controls (Custom Privacy function) are used to hide messages from unwanted targets, then direct message sharing is controlled, but information can still leak through mention propagation channels
Solution Approach 1:
The system performs preliminary analysis of the social network graph before message posting to identify all potential propagation paths including mention channels. By pre-computing the circle of trust that accounts for mention propagation, the system proactively prevents information leakage before it occurs, rather than reacting to it after posting.
Solution Approach 2:
The patent introduces an intermediary mechanism (the circle of trust computation system) that mediates between the user's posting action and the actual message propagation. This intermediary analyzes the network structure, identifies unwanted targets, and determines the safe circle of trust, thereby blocking harmful propagation paths while allowing legitimate sharing.
2Reliability
If users restrict sharing to a small group to avoid unwanted targets, then privacy is improved, but information dissemination to friends is limited
Solution Approach 1:
The circle of trust is dynamically computed based on the specific message, the user's social network structure, and the identified unwanted targets. Rather than using a static restricted group, the system adaptively determines the optimal sharing circle for each posting event, maximizing reach while maintaining privacy protection.
Solution Approach 2:
The system changes the parameter of trust evaluation from simple friend-list inclusion to a complex probability-based assessment that considers multiple propagation paths, mention channels, and unwanted target relationships. This parameter transformation enables more nuanced control that balances privacy and dissemination.
3Reliability
If comprehensive privacy controls tracking all propagation paths are implemented, then information leakage is prevented, but system complexity increases
Solution Approach 1:
The patent segments the complex propagation analysis into distinct components: identifying unwanted targets, computing propagation paths, calculating leakage probabilities, and determining the circle of trust. This segmentation makes the overall complex task manageable and implementable through modular algorithms.
Solution Approach 2:
The system automatically performs the complex analysis of propagation paths and circle of trust computation without requiring user intervention. Users simply specify unwanted targets, and the system self-services by computing the optimal sharing circle, eliminating the need for users to manually manage complex privacy settings.
4Reliability
If the circle of trust is constructed conservatively to ensure low leakage probability, then privacy is protected, but the number of friends who can receive information decreases
Solution Approach 1:
The system transforms the circle of trust from a static binary inclusion/exclusion list to a dynamic probabilistic construct. By changing parameters such as the maximum allowed leakage probability threshold, the system can flexibly adjust the circle size to balance privacy protection with information dissemination needs.
Solution Approach 2:
The system computes the circle of trust to achieve a specified leakage probability threshold, which may be more or less conservative than traditional approaches. By allowing partial inclusion based on probability assessments rather than absolute exclusion, the system optimizes the balance between privacy and reach.
Data Source
AI summary
Methods are provided for constructing an on-the-fly circle of trust for a user of an online social network to enable the user to reduce the likelihood that information will be leaked to an unwanted target. In one embodiment, a maximum circle of trust is constructed by using an iterative greedy construction (IGC) algorithm with leakage estimation using disjoint cut-sets. In another embodiment, the maximum circle of trust is constructed by using the IGC algorithm with leakage estimation using a hybrid method where the initial CT is constructed using the disjoint cut-sets and then the neighbors not included in the CT are sorted in non-decreasing order of visibilities and then each of these neighbors is included in the CT one at a time while checking if the leakage is below a threshold using a Sampling algorithm. In yet another embodiment, randomized rounding is used to construct the maximum circle of trust.