Machine Learning Model for Census-Level Audience Demographic Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for tracking digital media viewership, such as server logs and third-party cookies, are prone to over-counting and under-counting errors due to tampering, caching, and limitations in tracking cross-platform media exposure, making it challenging to accurately determine demographic classifications and unique audience sizes for digital advertising.
Innovation Solution
A system that utilizes machine learning models to analyze client device data, including media tags and demographic information from database proprietors, to deduplicate impressions and assign demographic classifications to census-level impression counts and unique audience sizes, without relying on third-party cookies, by reformating media tags to include additional features like DMA data and household demographics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If server logs and third-party cookies are used to track digital media viewership, then implementation is simple and cost-effective, but accuracy deteriorates due to over-counting and under-counting errors
Solution Approach 1:
The patent introduces a machine learning model as an intermediary between raw impression data and demographic classification. This model processes impression counts and unique audience sizes to predict demographic distributions, thereby improving measurement accuracy without requiring complex changes to the existing tracking infrastructure.
Solution Approach 2:
The patent transforms the tracking approach by changing from direct demographic measurement to probabilistic prediction. By using machine learning models that output probability distributions across demographic segments, the system achieves higher accuracy while maintaining implementation simplicity.
2Adaptability or versatility
If third-party cookies are used for cross-platform tracking, then audience measurement coverage is improved, but reliability deteriorates due to tampering and caching issues
Solution Approach 1:
The patent incorporates feedback mechanisms where the machine learning model is trained on known accurate data and continuously refined. The model learns from patterns in the data to compensate for tampering and caching issues, improving reliability while maintaining cross-platform coverage.
Solution Approach 2:
Instead of relying on vulnerable third-party cookies, the patent creates a parallel measurement system using machine learning models that replicate the functionality of direct tracking. This copying approach bypasses the reliability issues of cookie-based tracking while maintaining versatility.
3Measurement precision
If machine learning models are used to determine demographic classifications, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent applies machine learning models selectively to specific data processing tasks rather than attempting to redesign the entire tracking system. By focusing the ML application on demographic prediction only, the system achieves high precision without excessive overall complexity.
Solution Approach 2:
The machine learning model serves multiple functions: it classifies demographics, estimates unique audience sizes, and provides probability distributions for various segments. This multi-functionality reduces the need for separate systems, thereby limiting the increase in device complexity.
4Quantity of substance
If census-level impression counts are processed without demographic information, then data volume is reduced, but information completeness deteriorates
Solution Approach 1:
The patent replaces the mechanical approach of collecting detailed demographic data for every impression with a computational approach using machine learning. The model infers demographic information from aggregated census-level data, reducing data volume requirements while maintaining information completeness through probabilistic prediction.
Data Source
AI summary
Methods, apparatus, systems and articles of manufacture to determine demographic classifications for census level impression counts and unique audience sizes are disclosed. An example apparatus includes processor circuitry to execute instructions to execute a machine learning model to generate a probability of an occurrence of a demographic classification; generate a coviewing factor based on panel data; generate a viewer assignment output based on the coviewing factor; determine a unique audience total based on the probability of the occurrence of the demographic classification; adjust the unique audience total based on a non-coverage factor; and generate a report including the adjusted unique audience total.