Community Detection Algorithm for Impression Deduplication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge lies in accurately tracking media impressions across multiple devices, as existing methods using third-party cookies and digital ad identifiers are unreliable, leading to duplicated impressions and misrepresentation of true media exposure metrics due to shared email addresses and devices.
Innovation Solution
A network community monitor employs a community detection algorithm to deduplicate impressions by creating a device graph using PII-to-device links, grouping devices and users into clusters based on interaction frequency, and utilizing hyperparameters to optimize the objective function, thereby accurately attributing impressions to individual users.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If third-party cookies and digital ad identifiers are used to track media impressions, then impression tracking is enabled, but measurement precision deteriorates due to duplicated impressions and misrepresentation of true media exposure metrics
Solution Approach 1:
The patent introduces an intermediary system (audience measurement entity with device graph and community detection algorithm) that mediates between the tracking identifiers and the final measurement. This intermediary processes the raw tracking data through graph-based community detection to identify and resolve duplicate impressions, thereby improving measurement precision while maintaining the utility of impression tracking
Solution Approach 2:
The patent replaces the traditional mechanical approach of direct identifier-based tracking with a graph-theoretic computational system. Instead of relying on cookie and identifier matching, the system uses device graphs, community detection algorithms, and hyperparameter optimization to infer unique user exposure, substituting the mechanical tracking system with a more sophisticated computational measurement system
2Measurement precision
If devices and users are grouped into clusters based on interaction frequency, then duplicate impressions are eliminated, but device complexity increases due to the community detection algorithm and device graph construction
Solution Approach 1:
The patent segments the complex problem of duplicate impression resolution into manageable components: device graph construction, community detection algorithm execution, hyperparameter optimization, and impression deduplication. By dividing the system into these modular segments, the complexity is distributed and managed more effectively, allowing each component to be optimized independently
Solution Approach 2:
The patent performs preliminary actions by pre-construction of device graphs and pre-computation of community structures before actual impression measurement. This preliminary setup, including pre-defining hyperparameters and establishing device relationships, reduces the computational complexity during the actual measurement process, as the heavy lifting of graph construction and community detection is done in advance
Data Source
AI summary
Disclosed examples include at least one memory, instructions, and processor circuitry to execute the instructions to generate a device graph, the device graph to represent links between ones of personally identifiable information nodes and ones of device nodes, generate person-clusters based on the device graph, the person-clusters based on the links and community detection hyperparameter values, generate a node-to-person lookup structure based on the person-clusters, and deduplicate impression data based on the node-to-person lookup structure.