Hybrid Content Scheduling for Predicted Viewership Allocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing content scheduling systems struggle to effectively target users with content items based on their needs, interests, and locations, leading to inefficiencies and resource wastage, particularly when delivering addressable content.
Innovation Solution
A content scheduling system that predicts user viewing behaviors and optimizes the delivery of both addressable and non-addressable content items by clustering users into groups and selecting representative subsets for targeted delivery, maximizing effectiveness and minimizing resource waste.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If content items are delivered to all targeted users, then the quantity of content delivery increases, but resource wastage increases due to delivering to users who will not view the content
Solution Approach 1:
The system performs preliminary prediction of user viewing behavior before content delivery using machine learning models that analyze historical data, user profiles, and contextual factors. This preliminary action identifies which targeted users are likely to view content, allowing the system to pre-filter the delivery list and avoid wasting resources on users who will not engage with the content.
Solution Approach 2:
The system implements a feedback loop where actual viewing data from delivered content is fed back into the machine learning models to continuously improve prediction accuracy. This feedback mechanism allows the system to learn from past delivery outcomes and refine its targeting over time, reducing resource wastage while maintaining or increasing effective content delivery quantity.
2Reliability
If content delivery is targeted to specific users based on their characteristics, then the effectiveness of content delivery improves, but the complexity of the scheduling system increases
Solution Approach 1:
The system segments the user base into distinct groups or clusters based on shared characteristics, viewing behaviors, and preferences using unsupervised machine learning algorithms. This segmentation reduces complexity by handling groups rather than individual users, while still maintaining targeted effectiveness within each segment. The clustering approach creates manageable subsets that can be processed more efficiently.
Solution Approach 2:
The system introduces machine learning models as an intermediary layer between the content scheduling system and user characteristics. These models automatically process and interpret user data, transforming complex raw characteristics into simplified prediction scores or probability estimates. This intermediary handles the complexity of analyzing user characteristics, allowing the scheduling system to make targeted decisions based on model outputs rather than directly processing raw user data.
3Measurement precision
If the system predicts user viewing behavior for every content slot, then the precision of content placement improves, but the computational time and resources increase
Solution Approach 1:
The system applies partial prediction by focusing computational resources on predicting viewing behavior only for the most relevant content slots and user segments rather than every possible combination. Machine learning models prioritize predictions for high-value opportunities based on historical performance, contextual relevance, and potential impact, achieving sufficient precision for decision-making without the exhaustive computational burden of complete prediction.
Solution Approach 2:
The system performs preliminary aggregation and preprocessing of user data and content characteristics before the actual prediction process. By pre-computing relevant features, user profiles, and content attributes in advance, the system reduces the dimensionality and complexity of the prediction task. This preliminary action enables faster, more efficient predictions during content slot assignment while maintaining precision through the use of pre-processed, high-quality input data.
Data Source
AI summary
Systems, apparatuses, and methods are described for allocating content items for addressable content campaigns that target users with certain user characteristics and non-addressable content campaigns that request a certain quantity of deliveries in content delivery schedules. The insertions of addressable content items of the addressable content campaigns during slots in the content delivery schedules may be based on the predicted viewership (e.g., a number of views by targeted recipients) of the addressable content items during the slots. The deliveries of the content items from the addressable and non-addressable content campaigns may be scheduled to optimize values associated with viewings by targeted recipients of the assigned addressable content items and values associated with a quantity of deliveries of assigned non-addressable content items. The allocations of the contents items may also comply with various timing, geographical, delivery, slot inventory-related, and/or campaign specified constraints.