Machine Learning Demographic Classification for Census-Level Audiences
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for tracking digital media viewership, such as server logs and third-party cookies, are susceptible to over-counting and under-counting due to caching and privacy concerns, leading to inaccurate audience metrics and duplicate impressions across multiple devices.
Innovation Solution
A system utilizing a machine learning model that leverages database proprietor data and media tags to determine demographic classifications for census-level impression counts and unique audience sizes, without relying on third-party cookies, by using database proprietor aggregate subscriber-based audience metrics and encoding features like DMA data and household demographics to refine census-level impressions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If server logs and third-party cookies are used to track digital media viewership, then impression counts can be obtained, but over-counting and under-counting occur due to caching and privacy concerns
Solution Approach 1:
The patent introduces a machine learning model as an intermediary between raw impression data and demographic classification. This model processes census-level impression counts and unique audience sizes to infer demographic characteristics without relying on third-party cookies, thereby resolving the contradiction between measurement precision and reliability by using an intermediate computational layer that aggregates and analyzes data patterns
Solution Approach 2:
The patent creates a virtual representation of audience demographics through machine learning predictions rather than directly tracking individual users. By copying demographic patterns from trained model predictions on aggregated data, the system achieves accurate audience metrics without the reliability issues of direct tracking methods like third-party cookies
2Productivity
If third-party cookies are used for tracking, then audience metrics can be collected, but duplicate impressions across multiple devices are counted
Solution Approach 1:
The system uses database proprietor aggregate subscriber-based audience metrics that self-identify unique audiences through subscriber accounts rather than external tracking. Each subscriber's unique audience size is determined through their own authenticated sessions, eliminating duplicate counting across devices while maintaining productivity in metrics collection
Solution Approach 2:
The machine learning model acts as an intermediary that processes census-level impression counts and unique audience sizes to infer demographic characteristics. This intermediate layer aggregates data at the census level and uses pattern recognition to determine demographic classifications without directly tracking individual users across devices, thereby eliminating duplicate impression counting
3Quantity of substance
If census-level impression counts are used without demographic information, then impression volume is captured, but demographic classifications are unknown
Solution Approach 1:
The patent replaces traditional mechanical demographic collection methods (such as direct user profiling or cookie-based tracking) with a machine learning-based inference system. The model takes census-level impression counts as input and substitutes direct demographic measurement with predictive analysis, thereby recovering demographic information that would otherwise be lost while preserving the full volume of census-level data
Solution Approach 2:
The machine learning model is trained in advance on labeled demographic data to learn patterns and relationships between census-level metrics and demographic characteristics. This preliminary training action enables the model to infer demographic classifications from raw impression counts without requiring real-time demographic data collection, thus preventing information loss while maintaining full impression volume
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. In an example, the apparatus includes media tag format circuitry to generate a reformatted media tag corresponding to an impression request. The example apparatus also includes model execution circuitry to execute a machine learning model based on the reformatted media tag to generate outputs, the outputs including at least a value representative of a probability of an occurrence of a demographic classification. The example apparatus further includes audience counting circuitry to assign an identification of ones of audience members in a group to the demographic classification based at least on the outputs.