Distributed Transaction Scoring via Partial Score Aggregation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Decisioning systems face increased network bandwidth consumption and processing delays as the number of data feeds for scoring electronic transactions grows, necessitating a more efficient method to characterize financial transactions.
Innovation Solution
The system distributes the scoring process by polling multiple remote data sources for partial scores, which are then combined using methodologies like general linear models, neural networks, or support vector machines to generate a final score, allowing for scoring even before all data is received and leveraging remote processing resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the number of data feeds is increased to improve scoring accuracy, then measurement precision is improved, but network bandwidth consumption increases and processing delays occur
Solution Approach 1:
The patent segments the centralized scoring process into distributed partial scoring operations performed by multiple remote data sources. Each data source computes partial scores independently using provided scoring models, and these partial scores are aggregated to form the final transaction score. This segmentation reduces network bandwidth consumption by eliminating the need to transmit all raw data to a central system while maintaining scoring accuracy through comprehensive data source utilization.
2Measurement precision
If the number of data feeds is increased to improve scoring accuracy, then measurement precision is improved, but processing time increases
Solution Approach 1:
The patent applies preliminary action by pre-distributing scoring models to remote data sources before transaction scoring is needed. The models are stored locally at each data source, enabling them to immediately compute partial scores when transaction data arrives, eliminating the time required for model transmission and centralized processing. This preliminary preparation significantly reduces processing time while maintaining accurate scoring through comprehensive data source utilization.
3Reliability
If centralized scoring is used to ensure reliable score generation, then reliability is improved, but device complexity and processing requirements at the decisioning system increase
Solution Approach 1:
The patent implements self-service by enabling remote data sources to autonomously perform scoring operations using locally stored models and algorithms. Each data source independently computes its partial score without requiring centralized processing logic, reducing the complexity of the decisioning system. Reliability is maintained through the aggregation of multiple independent partial scores, which collectively provide a comprehensive and reliable transaction assessment.
Data Source
AI summary
A data transaction is associated with a plurality of remote data sources so that a score may be generated that characterizes the transaction. The remote data sources are polled to obtain partial scores which can be used by a decisioning system to score the transaction. In some implementations, different scoring models may be used so that the data source is notified which scoring model to use when generating the partial score. Related techniques, apparatus, systems, and articles are described.