Private Set Intersection for Cross-Regional User Graph Construction

The patent introduces an intermediary mechanism using homomorphic encryption and secret sharing protocols. Instead of directly exchanging user-identifiable information, the system uses encrypted representations of user data as intermediaries. These encrypted intermediaries allow for accurate graph construction through cryptographic operations without revealing actual user identities, thus resolving the contradiction between accuracy and privacy.

The patent transforms user-identifiable information into different parameter representations through encryption and anonymization. User IDs are converted into encrypted forms that maintain mathematical properties needed for graph construction while losing direct identifiability. This parameter transformation allows the system to work with modified data representations that satisfy both accuracy requirements and privacy constraints.

The patent segments the cryptographic operations into distinct modular components: homomorphic encryption module, secret sharing module, and graph construction module. Each component handles a specific aspect of the privacy-preserving process independently. This segmentation reduces system complexity by making each module manageable and reusable, while collectively achieving strong privacy protection.

The system implements self-service mechanisms where the cryptographic protocols automatically perform privacy protection without requiring manual intervention. The homomorphic encryption and secret sharing operations are integrated into the graph construction workflow, allowing the system to autonomously protect data privacy while constructing the user-user graph, thus reducing operational complexity.

The patent replaces traditional mechanical identifier matching with cryptographic operations. Instead of directly comparing user IDs to build graph edges, the system uses homomorphic encryption to perform computations on encrypted identifiers. This substitution allows graph analysis to be conducted on cryptographic representations that preserve privacy while maintaining the mathematical properties necessary for accurate analysis.

The system creates a composite data structure combining multiple properties: encrypted user identifiers, anonymized node representations, and cryptographic proofs. This composite structure integrates both privacy protection mechanisms and analysis-enabling features, allowing the graph to simultaneously maintain data privacy and support accurate graph analysis through its multi-layered design.