Systems and methods of log optimization for television advertisements
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current log optimizers for television advertisements face complexity due to thousands of constraints, leading to suboptimal revenue outcomes as they often settle for local maxima instead of global maxima, missing significant revenue potential.
Innovation Solution
Utilizing linear integer programming (LIP) optimization techniques with graph theory processes to reduce constraints, combined with machine learning for viewership predictions, to generate an optimized advertisement schedule that maximizes revenue yield.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If greedy optimization techniques are used to handle thousands of ads and constraints, then the optimization process can proceed sequentially with manageable complexity, but the solution converges to local maxima and misses significant revenue potential
Solution Approach 1:
The patent segments the optimization problem into two distinct phases: a preprocessing phase that performs graph theory operations to reduce constraints and identify swap opportunities, and a subsequent optimization phase that evaluates and executes swaps. This segmentation allows the system to handle the complexity of thousands of ads and constraints while avoiding local maxima by considering multiple simultaneous swaps rather than sequential greedy choices.
2Productivity
If all possible simultaneous swaps are inspected to find the global maximum solution, then the highest possible predicted log value can be achieved, but the problem becomes intractable for current greedy optimizers
Solution Approach 1:
The patent applies preliminary action by performing graph theory preprocessing before the main optimization step. This preprocessing identifies potential swap opportunities and reduces the constraint set, so that when the optimization phase evaluates simultaneous swaps, the computational burden is already reduced. This allows the system to approach the global maximum without facing the full intractability of inspecting all possible swaps from scratch.
3Ease of operation
If greedy optimizers perform sequential swaps to gradually increase predicted delivery, then the process is computationally manageable, but the optimizer stops at local maxima and cannot achieve higher revenue
Solution Approach 1:
The patent transitions from the traditional sequential one-dimensional optimization approach to evaluating multiple simultaneous swaps across different dimensions of the advertisement log. By using graph theory to identify and evaluate multiple swap opportunities in parallel rather than sequentially, the system can escape local maxima and achieve higher predicted log values while maintaining computational feasibility through smart preprocessing.
Data Source
AI summary
Embodiments of the present invention provide systems and methods of log optimization for television advertisements. Exemplary method and systems can comprise: receiving, at a computer software platform, (i) the digital log, (ii) constraint parameters, and (iii) viewership predictions; and generating, with the computer software platform, an optimized log based on the received digital log, the constraint parameters, and the viewership predictions.