Audience Metrics Estimator Using Probability Divergence for Census Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current audience measurement systems face challenges in accurately estimating census-level total impression durations and audience size across demographics due to anonymity in census-level impressions, particularly when third-party data provides partial information on durations and unique audience sizes.
Innovation Solution
The implementation of an audience metrics estimator that utilizes information theory to determine probabilities of individuals being part of third-party subscriber data, calculates probability divergences, and establishes a search space within bounds based on equality constraints to estimate census-level unique audience and durations across demographics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If census-level impressions are used to measure audience size and duration, then anonymity and privacy are preserved, but measurement precision and demographic accuracy deteriorate
Solution Approach 1:
The patent introduces a probabilistic modeling intermediary that bridges census-level anonymous impressions and third-party demographic data. The system uses probability distributions and divergence calculations as mediators to infer demographic characteristics without directly linking anonymous impressions to specific individuals, thus preserving anonymity while improving measurement precision
Solution Approach 2:
The system transforms the measurement approach by changing from direct demographic attribution to probabilistic parameter estimation. By using probability distributions over demographic parameters and calculating divergences between observed and expected distributions, the system achieves accurate demographic measurement while maintaining census-level anonymity
2Measurement precision
If third-party subscriber data is used to improve demographic accuracy, then measurement precision improves, but data completeness and reliability deteriorate due to partial information
Solution Approach 1:
The patent applies partial action by utilizing only the portions of third-party data that are reliable and complete, while using probabilistic methods to estimate the missing portions. The system calculates probability distributions based on available third-party subscriber data and uses divergence measurements to infer characteristics of non-subscribers, achieving complete demographic coverage without requiring complete third-party data
Solution Approach 2:
The system implements feedback through iterative probabilistic modeling where initial estimates are refined by comparing observed census-level impression patterns against predictions from third-party data. The probability divergence calculations provide feedback signals that adjust the model parameters to better align with actual observed data, improving both precision and reliability
Data Source
AI summary
Example technical solutions to determine census-level audience metrics across demographics are disclosed. Disclosed example apparatus include a distribution parameter solver to initialize distribution parameter values for a probability of an individual within a demographic being (i) included in a subscriber audience for the demographic and (ii) having a first impression duration, the subscriber audience having a first subscriber audience size. Disclosed example apparatus also include a divergence parameter solver to determine divergence parameter values between (i) the first subscriber audience size and the first impression duration and (ii) a census-level audience size and a second impression duration based on the initialized distribution parameter values. Disclosed example apparatus further include a search space identifier to identify a search space within bounds based on a census-level total impression count and a census-level total impression duration, the search space to define a range based on census-level duration equality and census-level audience equality constraints.