Separate Models Adjust Representation Parameters for Content Competition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The dynamic and unpredictable nature of user account characteristics and line item competition in video delivery systems makes it difficult to estimate the delivery of supplemental content, such as advertisements, leading to challenges in setting delivery goals and targeting requirements.
Innovation Solution
A system utilizing two separate models to generate representation parameters for estimating the delivery effectiveness of supplemental content, where a first model generates estimated values based on line item-specific information and a second model applies modification factors to adjust these values based on competition factors, improving training efficiency and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single model is used to estimate delivery parameters, then the model complexity is low, but the accuracy of estimating supplemental content delivery is insufficient due to dynamic user characteristics and line item competition
Solution Approach 1:
The patent divides the delivery estimation model into two separate models: a first model that estimates delivery parameters based on line item factors, and a second model that generates competition factors based on line item competition. This segmentation allows each model to specialize in specific aspects of the estimation problem, improving overall accuracy while managing complexity through modular design.
Solution Approach 2:
The patent introduces competition factors as an intermediary element that mediates between line item characteristics and delivery estimation. The second model generates these competition factors that capture the dynamic competitive environment, which then influence the final delivery parameter estimates from the first model, enabling more accurate representation of market dynamics.
2Measurement precision
If historical data is required for model training, then the accuracy of delivery estimation improves, but the system cannot effectively handle new line items without historical data
Solution Approach 1:
The patent performs preliminary actions by training the first model on historical delivery data and the second model on historical competition data before deployment. This preliminary training establishes baseline capabilities that allow the models to make reasonable estimates for new line items even without specific historical data for those items, while still adapting to new situations.
Solution Approach 2:
The patent enables parameter changes by allowing the competition factors generated by the second model to dynamically adjust the delivery parameters from the first model. This mechanism allows the system to adapt to new line items and changing market conditions by modifying the estimated delivery parameters based on current competition dynamics, even when historical data for specific new line items is unavailable.
Data Source
AI summary
In some embodiments, a method receives a line item factor for a line item at a first model. The method receives a competition factor at a second model. The first model generates a first value for a representation parameter based on the line item factor. The representation parameter is used to generate a representation for estimating selection of the instance of supplemental content for the line item for delivery. The second model generates a modification factor based on the competition factor. The modification factor is configured to modify the first value for the representation parameter based on competition from other line items to select the instance of supplemental content for the line item for delivery. The method modifies the first value for the representation parameter using the modification factor to generate a second value for the representation parameter where the second value is used to generate the representation.