Prediction Network for Immersive Video Content Delivery
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Video delivery systems face challenges in providing an optimal viewing experience for users exhibiting immersive viewing behavior, as traditional methods of delivering supplemental content can disrupt the main content and negatively impact user perception, especially for users engaged in binge watching multiple episodes of the same show.
Innovation Solution
A video delivery system that predicts user behavior using a combination of short-term and long-term viewing patterns, employing a prediction network with spatial locality and attention-based mechanisms to deliver supplemental content in a format that minimizes interruptions during immersive viewing sessions, such as delivering it before the main content starts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If supplemental content is delivered during breaks of main content for all user accounts, then the system can provide advertising revenue and engage casual viewers, but immersive viewers experience interruptions that negatively affect their perception of the service
Solution Approach 1:
The system predicts immersive viewing behavior in advance using historical data and machine learning models. When immersion is predicted, supplemental content delivery is preemptively adjusted by delivering it before or after the main content rather than during breaks, thus preventing interruptions before they occur and preserving the immersive experience.
2Ease of operation
If traditional fixed timing breaks are used for all users, then the system operation is simple and consistent, but it fails to adapt to different viewing behaviors and reduces engagement for immersive viewers
Solution Approach 1:
The system transitions from static fixed timing breaks to dynamic adaptive scheduling. Machine learning models continuously analyze user behavior patterns and adjust break timing and supplemental content delivery in real-time based on detected immersion levels, allowing the system to adapt its operation to individual viewing behaviors while maintaining simplicity through automated decision-making.
Solution Approach 2:
The system changes key parameters such as break timing, supplemental content delivery timing, and content selection based on detected viewing behavior. When immersion is detected, parameters are adjusted to delay or relocate supplemental content delivery, transforming the rigid fixed-timing approach into a flexible behavior-responsive system.
3Quantity of substance
If interruptions are provided to all users during main content, then the system can maximize supplemental content exposure, but immersive viewers become less responsive and their perception of the service deteriorates
Solution Approach 1:
The system applies different supplemental content delivery strategies to different user segments based on their viewing behavior. Immersive viewers receive supplemental content at different times (before or after main content) compared to casual viewers (during breaks), allowing the system to maintain high supplemental content exposure overall while preserving responsiveness and service perception for the immersive segment.
Data Source
AI summary
In some embodiments, a method inputs a sequence of historical behaviors for a plurality of instances of content into a prediction network to generate a sequence of values that model the sequence of historical behaviors. A restriction on an operation performed by the prediction network is based on a characteristic of an viewing behavior. A sequence of attention scores is generated based on a similarity of a current behavior for a first instance of content to respective instances of historical behaviors in the sequence of historical behaviors. The method adjusts respective values based on corresponding attention scores to generate an adjusted sequence of values. The adjusted sequence of features are sampled to generate an output from the prediction network that models the sequence of historical behaviors based on the current behavior. The output for determining a prediction if the current behavior is indicative of the viewing behavior.