Entity Resolution Over-Match Correction via Minimum Cut Pathway Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Entity resolution techniques often result in over-matching, where records that do not describe the same entity are incorrectly linked due to flexible matching criteria, making it difficult to distinguish indirect matches and consider non-matching data.
Innovation Solution
A computer-program product that generates a record indication by setting peripheral records as a source and sink, using a minimum cut algorithm to identify pathways and compute metrics for linked records, allowing for the disconnection of incorrect linkages and separation of entities based on record linkage information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If flexible matching criteria are used in entity resolution to link records, then more records can be matched and resolved, but over-matching occurs where records that do not describe the same entity are incorrectly linked
Solution Approach 1:
The patent segments the entity resolution process into multiple stages: initial flexible matching to identify candidate records, followed by pathway analysis that divides the matching process into distinct connection paths. Each pathway is evaluated separately using minimum cut algorithms to determine if it represents a valid entity linkage or an over-match error, allowing the system to maintain high productivity while improving precision through structured analysis of matching pathways
Solution Approach 2:
The patent introduces pathway analysis as an intermediary layer between flexible matching and final entity resolution. This intermediary process uses minimum cut algorithms to evaluate matching pathways and identify over-matches, acting as a mediator that filters incorrect linkages while preserving valid matches, thus resolving the contradiction between high matching volume and accuracy
2Measurement precision
If multiple pathways are analyzed using minimum cut algorithm to identify over-matches, then accuracy of entity resolution improves, but computational complexity increases
Solution Approach 1:
The patent applies preliminary action by first identifying candidate records using flexible matching criteria before conducting pathway analysis. This preliminary filtering step reduces the number of records that require complex minimum cut algorithm analysis, allowing the system to achieve high accuracy through pathway evaluation while managing computational complexity by limiting detailed analysis to only candidate records that pass the initial matching stage
Solution Approach 2:
The patent implements partial action by applying the computationally intensive minimum cut algorithm selectively rather than to all possible record combinations. The system identifies over-matches by analyzing pathways only for candidate records that meet initial matching criteria, performing partial pathway analysis sufficient to detect over-matches without exhaustively evaluating all possible match scenarios, thus balancing accuracy with computational feasibility
Data Source
AI summary
A computing device obtains an indication of data records resolved to describe a single entity in an entity resolution. The data records comprise peripheral records resolved to describe the single entity based on matching data of a central record of the data records. The device generates an indication indicating that at least one of the first peripheral record and the second peripheral record does not describe the single entity by setting a first one of the data records as a source; and setting a second one of the data records as a sink. The device generates a data structure identifying record linkage information for records of the dataset. The record linkage information indicates one or more pathways between the source and the sink along the linked records. The device executes a minimum cut algorithm to identify one or more connections of the one or more pathways to unlink.