Record Linkage via Behavior Recognition Gain

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing record linkage techniques face challenges in accurately matching data records from multiple entities with incomplete knowledge of entity behavior, as similarity measurements can group similar behaviors together but fail to find unique matches, and are misleading when comparing partial behaviors.

Innovation Solution

A computer-implemented method that uses transaction logs to determine a matching score by generating a measure representing a gain in behavior recognition before and after merging entities, employing a statistical model to calculate a final matching score and transforming logs into a predetermined format to extract behavior data for coarse and fine matches.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If similarity measurement is used to match data records from multiple entities, then entities with similar behaviors can be grouped together, but unique matches cannot be identified and partial behavior comparisons become misleading

Engineering Contradiction:

Improvematching accuracyVSAvoidincomplete behavior knowledge

Core Design Contradiction:

Measurement precisionVSLoss of information

Solution Approach 1:

The patent merges behavior data from multiple sources for the same entity to create a complete behavior profile. By combining transaction logs from different data sources, the system overcomes the limitation of partial behavior knowledge and enables accurate unique matching rather than just grouping similar entities.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

Instead of comparing behaviors to find similarities, the patent inverts the approach by measuring the gain in behavior recognition when merging entities. The matching score is based on how much the combined behavior profile improves recognition, rather than how similar the individual behaviors are.

Inventive Principle:

Principle #13The other way round (Inversion)

2Reliability

If complete behavior knowledge is available for both sources, then similarity comparison can be performed, but entities with very similar behaviors cannot be distinguished for unique matching

Engineering Contradiction:

Improvebehavior comparison reliabilityVSAvoidunique match identification

Core Design Contradiction:

ReliabilityVSMeasurement precision

Solution Approach 1:

The patent fundamentally inverts the matching approach by not measuring similarity between entities, but rather measuring the gain in behavior recognition achieved by merging them. This inversion allows unique identification even when behaviors are very similar, because the merge gain captures subtle complementary information.

Inventive Principle:

Principle #13The other way round (Inversion)

Solution Approach 2:

The patent introduces an intermediary concept of 'behavior recognition gain' that mediates between similar behaviors. This intermediate metric captures the incremental information value of merging, enabling discrimination between entities with otherwise similar behaviors.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Adaptability or versatility

If traditional similarity strategies are used, then entities with similar behaviors are grouped together, but the process cannot identify unique matches across different data sources

Engineering Contradiction:

Improvebehavior grouping capabilityVSAvoidunique entity identification

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent inverts the traditional similarity-based grouping approach by using behavior recognition gain from merging. Instead of asking 'how similar are these entities?', it asks 'how much does merging these entities improve behavior recognition?', enabling unique identification while maintaining adaptability.

Inventive Principle:

Principle #13The other way round (Inversion)

Solution Approach 2:

The patent changes the fundamental parameter used for matching from similarity measure to behavior recognition gain. This parameter change transforms the matching objective from grouping similar entities to identifying unique matches through the incremental value of merged behavior profiles.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS9514167B2Behavior based record linkage

Publication Date: 2016.12.06 HAMAD BIN KHALIFA UNIVERSITY

AI summary

A computer implemented method for matching data records from multiple entities comprising providing respective transaction logs for the entities representing actions performed by or in respect of the entities, determining a matching score using the transaction logs for respective pairs of the entities and for predetermined combinations of merged entities by generating a measure representing a gain in behavior recognition for the entities before and after merging, and using the gain as a matching score.