MAID Graph Weighting for Multi-Person Device Identity Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for matching Mobile Ad Identifiers (MAIDs) with Personal Identifiers (PIDs) fail to accurately associate multiple individuals using a common device, leading to inefficiencies and computational challenges, particularly in probabilistic modeling approaches.
Innovation Solution
A data graph structure is constructed using machine learning to rank PIDs by likelihood of use, assigning weights to MAIDs, reducing computational complexity by converting raw data into single numerical scores, and utilizing deterministic associations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If probabilistic methods are used to model which person might be using a CTV device, then the system can handle multiple persons using a common device, but the computational complexity increases significantly
Solution Approach 1:
The patent uses simple, deterministic rules instead of complex probabilistic models. Each MAID is associated with a single PID through straightforward matching criteria (device type, operating system, user profile), eliminating the need for computationally intensive probabilistic calculations while still handling multiple persons per device scenarios.
Solution Approach 2:
The patent segments the problem by creating separate MAID-PID associations for different device types and contexts. Rather than using a single complex probabilistic model for all devices, it divides the solution into device-specific deterministic rules, reducing overall computational complexity while maintaining versatility.
2Device complexity
If a single PID is assigned to each MAID representing a single person, then the data processing is simple, but the system fails to address multiple persons using a common device
Solution Approach 1:
The patent makes the MAID-PID association system universal by allowing multiple PIDs to be associated with a single MAID through deterministic rules. The same matching logic applies across different device types (mobile phones, tablets, CTVs), and the system automatically adapts to handle both single-person and multi-person scenarios without requiring different approaches.
3Measurement precision
If probabilistic modeling is used to determine user identity, then the system can account for uncertainty in device usage, but the computational requirements become impractical (Terabytes of data)
Solution Approach 1:
The patent enables the system to determine user identity through self-service deterministic matching without requiring external probabilistic modeling. The MAID-PID associations are established through direct matching of device characteristics and user profiles, allowing the system to achieve accurate identification with minimal data processing requirements.
Data Source
AI summary
A system for matching personal identifiers (PIDs) to Mobile Ad Identifiers (MAIDs) allows multiple PIDs to be associated with each MAID by constructing a data graph structure, with weights assigned to representing the likelihood of each of the persons associated with each of the PIDs to be using the device associated with a MAID at a given time. The graph is constructed by a machine learning model that ranks PIDs by which is most likely to be seen in the near future using the device to which the MAID is associated. These results are then converted into probabilities. MAIDs are represented as nodes in the graph, while probabilities are stored as edges.