System for the data protection-compliant joint analysis of fraud-relevant transaction patterns in distributed computing environments

Abstract

System for the data protection-compliant joint analysis of fraud-relevant transaction patterns in distributed computing environments, comprehensive In every computing environment, a local processing unit is trained to capture transaction data and to derive fraud-relevant pattern characteristics from the transaction data, and a coordination unit, trained to align the analysis processes between several separate computing environments, wherein only derived analysis information is exchanged between the computing environments, while the underlying transaction data remains in the respective computing environment, and wherein the coordination unit performs a state-dependent feedback control by which the local processing units iteratively adjust their derivation of the pattern features without a central data collection or central pattern calculation.

Description

Technical field

[0001] The invention relates to the technical field of distributed computing systems and security-related data processing. In particular, the invention relates to a technical system for the joint analysis of fraud-relevant transaction patterns in several separate computing environments while maintaining data confidentiality. State of the art

[0002] In distributed computing environments, fraud-relevant transaction data is typically either evaluated locally in isolation or centrally aggregated to perform pattern recognition. However, central aggregation requires the transfer of sensitive data, leading to increased data privacy risks, regulatory restrictions, and dependencies on central evaluation centers.

[0003] Known distributed evaluation methods are often limited to the exchange of aggregated results or static parameters and do not allow coordinated, iterative pattern analysis across multiple independent computing environments.

[0004] A system that enables a joint analysis of fraud-relevant transaction patterns without the underlying transaction data leaving the respective computing environment is not known from the prior art. Object of the invention

[0005] The invention is based on the objective of providing a technical system that: • a joint analysis of fraud-relevant transaction patterns enables, • avoids the transmission of sensitive transaction data, • can be used in multiple independent computing environments, and • supports data protection-compliant collaboration between computing environments. Summary of the invention

[0006] The invention relates to a system for the data protection-compliant collaborative analysis of fraud-relevant transaction patterns in distributed computing environments. Each computing environment contains a local processing unit that captures transaction data and derives fraud-relevant pattern characteristics from it. A coordination unit synchronizes the analysis processes between several separate computing environments, whereby only derived analysis information is exchanged and the underlying transaction data remains within the respective computing environment. The coordination unit performs state-dependent feedback control, through which the local processing units iteratively adjust their derivation of pattern characteristics without a central data collection or pattern calculation.The joint analysis is performed in a time-segmented and asynchronous manner, enabling a distributed, continuous and privacy-compliant pattern analysis without time synchronization of the computing environments.

[0007] The system includes a local analysis unit in each computing environment for identifying fraud-relevant pattern characteristics, as well as a coordination unit for aligning the analysis processes between the computing environments. Only derived analysis information is exchanged between the computing environments, while the underlying transaction data remains within the respective computing environment. Detailed description of the invention

[0008] The invention relates to a system for the data protection-compliant joint analysis of fraud-relevant transaction patterns in distributed computing environments. Each computing environment contains a local processing unit for deriving fraud-relevant pattern characteristics from transaction data. A coordination unit enables the alignment of the analysis processes between the computing environments, whereby only derived analysis information is exchanged. The underlying transaction data remains within the respective computing environment. This enables a distributed, coordinated, and data protection-compliant analysis of fraud-relevant transaction patterns.

[0009] Each participating computing environment includes a local processing unit that captures transaction data and derives fraud-relevant pattern characteristics from it. These pattern characteristics represent abstract properties of transaction flows without revealing individual transactions or personal data.

[0010] A coordination unit is trained to align the local analysis processes between the computing environments. For this purpose, only structured analysis results or adjustment information are transferred between the computing environments.

[0011] Based on the coordinated analysis, the local processing units can adapt their pattern recognition without requiring a central data collection system. The joint analysis is thus distributed, coordinated, and compliant with data protection regulations.

Claims

[1] System for data protection-compliant joint analysis of fraud-relevant transaction patterns in distributed computing environments, comprising In every computing environment, a local processing unit is trained to capture transaction data and to derive fraud-relevant pattern characteristics from the transaction data, and a coordination unit, trained to align the analysis processes between several separate computing environments, wherein only derived analysis information is exchanged between the computing environments, while the underlying transaction data remains in the respective computing environment, and wherein the coordination unit performs a state-dependent feedback control by which the local processing units iteratively adjust their derivation of the pattern features without a central data collection or central pattern calculation. [2] System according to claim 1, wherein the coordination unit is configured to generate consistency-related adaptation information that describes differences between the pattern features derived in the computing environments and to transmit this adaptation information to the local processing units, so that a convergent but decentrally executed pattern analysis is achieved across multiple computing environments. [3] System according to claim 1, wherein the local processing units derive the fraud-relevant pattern features in a temporally segmented manner and the coordination unit provides the adaptation information section by section and asynchronously, so that the joint analysis is carried out without temporal synchronization of the computing environments and a scalable, continuous pattern adaptation is realized.