Granted Patent B2

US 12,596,692 · App. 18/019,736 · Granted Apr 7, 2026

Inventors: Dwayne Collins (Conway, AR); Pavan Roy Marupally (Conway, AR)

Assignee: LiveRamp, Inc.

G06F16/215G06F16/288G06F16/9024

Abstract

Claims (22)

1 . A computerized method for performing source valuation within a parallel processing environment for a data source considered for inclusion in an entity resolution data graph, the method comprising the steps of:

using a plurality of processors in a parallel processing computer system to:

construct a first sample of the data source, wherein the first sample comprises a first subset of records in the data source associated with a dense representation of a geographically localized area;

construct a second sample of the data source, wherein the second sample comprises a second subset of records randomly selected from the data source, excluding the first subset of records;

automatically partition the first sample and second sample into a plurality of partitions using a partitioning algorithm that distributes data across multiple processors based on processing capacity, wherein each partition corresponds to an available processor within the parallel processing environment;

automatically identify corresponding relationships between sets of fields in the data source and nodes in the entity resolution data graph using automated field mapping algorithms that compare data structures and analyze entity relationship patterns;

execute in parallel across the plurality of processors a profiling process that analyzes one or more fields and the identified sets of fields to automatically generate sets of distributions, sets of counts, and sets of examples into a machine readable profiling scorecard that characterizes entity resolution quality metrics;

automatically aggregate the sets of distributions, set of counts, and set of examples into a machine-readable profiling scorecard that integrates entity resolution confidence scores with data quality assessments;

automatically compare the set of distributions, set of counts, and set of examples to at least one authoritative source distribution using statistical comparison algorithms to produce a quality score for determining data source trustworthiness, wherein the quality score combines data profiling results with entity matching validation;

pass the first sample and second sample into a match service, wherein the match service uses an identity graph comprising unique person identifiers, address identifiers, and household identifiers, and executes cascade level matching at a plurality of cascade levels;

return from the match service, for each record in the first sample and second sample, person identifiers for each of the plurality of cascade levels;

construct a key tuple for each record, wherein the key tuple comprises a plurality of sub-tuples each corresponding to a cascade level and containing person identifiers returned from the match service for that cascade level, wherein a first positive integer value indicates a known person in the entity resolution data graph, wherein different positive integer values for different cascade levels within a single record indicate that persons matched at different cascade levels may be the same person and represent a potential consolidation, and wherein a negative value indicates that a touchpoint is not in the entity resolution data graph and represents new information;

aggregate the key tuples into a distribution document whose values are counts of records sharing the same key; and

determine, from the distribution document, a level of trustworthiness of the data source and an expected impact on the entity resolution data graph if the data source is added, wherein the expected impact indicates at least one of: new touchpoints to be added, new relationships for known touchpoints, or potential consolidation of persons.

2 . The method of claim 1 , wherein the first subset of records comprises about two percent of a total number of records in the data source.

3 . The method of claim 1 , wherein the second subset of records comprises about three percent of a total number of records in the data source.

4 . The method of claim 1 , wherein the sets of fields comprise postal state and phone area code.

5 . The method of claim 1 , wherein the sets of fields comprise gender and age.

6 . The method of claim 1 , wherein the authoritative source distribution comprises US Census data.

7 . The method of claim 1 , wherein the authoritative source distribution comprises US Postal Service data.

8 . The method of claim 1 , wherein the plurality of matched cascade levels comprises a postal address only level.

9 . The method of claim 1 , wherein the plurality of matched cascade levels comprises a name plus telephone number level.

Assignments (1)

Continuity (2)

References Cited (23)