Graph Circuit Analysis for Identity Resolution Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods fail to accurately measure the accuracy of identity resolution in large databases, such as marketing databases, which contain hundreds of millions or billions of data elements, leading to inefficiencies and potential miscorrelation of data elements.
Innovation Solution
Applying electrical circuit analysis techniques by constructing a graph with nodes and edges representing data elements, where edge strength is proportional to the likelihood of elements pertaining to the same object, using series and parallel connections to determine accuracy through Kirchhoff circuit analysis laws.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional identity resolution methods are used in large databases, then data correlation can be performed, but measurement accuracy of resolution results cannot be determined and computational efficiency is insufficient
Solution Approach 1:
The patent replaces traditional statistical and machine learning-based identity resolution methods with an electrical circuit analysis model. Data elements are represented as electrical nodes, relationships as conductive paths with specific conductances, and resolution accuracy is determined through circuit laws (Ohm's law, Kirchhoff's laws). This substitution enables precise measurement of resolution accuracy through electrical potential differences while maintaining computational efficiency through circuit simulation techniques.
Solution Approach 2:
The patent transforms the identity resolution problem into an electrical circuit problem by changing the parameter representation: data element relationships are represented as conductance values, confidence scores as electrical potentials, and resolution accuracy as measurable voltage differences. This parameter transformation enables the application of well-established circuit analysis methods to measure and optimize identity resolution accuracy in large databases.
2Reliability
If comprehensive data correlation is performed across billions of data elements, then resolution coverage is improved, but computational time becomes impractical
Solution Approach 1:
The patent segments the large-scale identity resolution problem into smaller electrical circuit sub-problems that can be solved independently and then combined. The database is divided into multiple data centers or partitions, each represented as a separate circuit network. Circuit analysis can be performed on these segmented networks in parallel, significantly reducing computational time while maintaining comprehensive resolution coverage across all billions of data elements.
Solution Approach 2:
The patent introduces electrical circuit simulation as an intermediary layer between the raw data and the identity resolution results. Instead of directly comparing all data element pairs, the system uses circuit analysis algorithms as intermediaries to efficiently compute relationships across billions of elements. This intermediary approach enables comprehensive resolution coverage through the use of optimized circuit simulation techniques that scale better than traditional pairwise comparison methods.
Data Source
AI summary
A system and method utilizes a data integration input routine receive raw data set(s) from identity data storage media resources, generate an edge type from each data set, and store the edge type from each data set in a first temporary storage media, from which a graph construction module retrieves the edge types and combines them to produce a consolidated edge store, a search of which is used to find graph component paths. Current paths are joined against the consolidated edge store to find edges that extend each path in the consolidated edge store, those paths that extend are stored in a graph component table, from which a sample of graph paths are downloaded and a graph is constructed. A circuit analysis engine is used to perform a circuit analysis and a selectivity module is used to selectively modify the scope of the circuit analysis and results.