Audience Deduplication Using Count Sketch Vectors and Binomial Hashing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audience measurement systems face challenges in accurately estimating audience sizes across multiple platforms due to duplicate counting of media exposures and the inability to share personal identifiable information (PII) while maintaining privacy, especially with the decline of third-party cookies.
Innovation Solution
The use of sketch data generated by database proprietors using binomial hashing to create vectors of counts that summarize media exposure without revealing PII, combined with a method to estimate overlap between datasets to deduplicate audience members across different platforms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional panel member monitoring is used, then measurement precision is maintained, but coverage is limited to enrolled members only
Solution Approach 1:
The patent merges traditional panel member data with third-party cookie data from multiple database proprietors into a unified audience measurement system. This combination allows the system to maintain measurement precision for panel members while extending coverage to include non-panel members accessed through various digital platforms, thereby resolving the contradiction between measurement accuracy and platform coverage.
Solution Approach 2:
The patent introduces an intermediary deduplication process that uses sketch data and binomial hashing to bridge panel member data with third-party cookie data. This intermediary mechanism enables the system to integrate data from multiple sources without directly sharing PII, allowing expansion of coverage while maintaining measurement precision through controlled data fusion.
2Adaptability or versatility
If third-party cookies are used for tracking, then platform coverage is expanded, but privacy is compromised
Solution Approach 1:
The patent extracts only the necessary tracking functionality from third-party cookies while discarding the privacy-invasive aspects. By using sketch data and binomial hashing, the system extracts audience measurement capability without extracting or storing PII, thereby maintaining cross-platform tracking capability while eliminating privacy violations.
Solution Approach 2:
The patent creates a copy of audience identification through sketch data representations rather than using direct PII from third-party cookies. This copying approach allows the system to maintain cross-platform tracking capability through anonymized data representations while avoiding the privacy harms associated with direct PII usage.
3Adaptability or versatility
If data from multiple database proprietors is combined, then coverage is improved, but duplicate counting increases
Solution Approach 1:
The patent implements a feedback mechanism through the deduplication process that continuously adjusts the audience count based on overlap detection between database proprietor datasets. By using sketch data comparisons, the system receives feedback about duplicate entries and corrects them, thereby improving measurement precision while maintaining expanded coverage.
Solution Approach 2:
The patent replaces traditional mechanical deduplication methods with a probabilistic approach using binomial hashing and sketch data. This substitution allows for more efficient and accurate duplicate detection across multiple datasets, improving the accuracy of unique audience size measurement while maintaining comprehensive multi-platform coverage.
4Measurement precision
If PII is shared between database proprietors, then deduplication accuracy is improved, but privacy protection is weakened
Solution Approach 1:
The patent introduces sketch data and binomial hashing as intermediary mechanisms that enable deduplication accuracy without requiring direct PII sharing between database proprietors. These intermediaries facilitate the measurement process while maintaining privacy protection, resolving the contradiction between deduplication accuracy and privacy risk.
Solution Approach 2:
The patent changes the parameters of data representation from direct PII to anonymized sketch data with binomial hash values. This parameter transformation maintains the necessary information for accurate deduplication while eliminating privacy risks associated with PII sharing, thereby resolving the contradiction between measurement precision and privacy protection.
Data Source
AI summary
Methods and apparatus to estimate audience sizes using deduplication based on vector of counts sketch data are disclosed. A system includes hardware circuitry to instantiate: coefficient analyzer circuitry to determine coefficient values of a polynomial based on (i) variances in values in a first vector of counts and a second vector of counts, (ii) a first cardinality of the first vector of counts, and (iii) a second cardinality of the second vector of counts; and overlap analyzer circuitry to: determine a real root of the polynomial; and report generator circuitry to estimate a deduplicated audience size based on (i) the estimate of the quantity of the second subscribers that are duplicates of the first subscribers and (ii) the first and second cardinalities. The system includes communication circuitry to transmit a network communication to a third party entity, the second network communication including a report based on the deduplicated audience size.