Dynamic Watch List Reordering via User Behavior Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video delivery systems lack a dynamic and personalized method to order recurring episodes of shows based on user watch history, leading to static lists that do not accurately predict user preferences.
Innovation Solution
A method using a machine learning predictor to analyze user watch history and determine selection probabilities for episodes, dynamically updating a watch list by categorizing shows based on their status and release timing, and grouping them for accurate prediction of user viewing order.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a static list of videos is used (e.g., DVR list or on-demand queue), then the system is simple to implement, but the list does not dynamically adapt to user preferences and behavior patterns
Solution Approach 1:
The patent transforms the static video list into a dynamic system that automatically updates based on user behavior. The watch list dynamically reorders videos by predicting user selection probability, changing the list structure in real-time as users interact with content, thereby achieving adaptability without manual intervention
Solution Approach 2:
The system implements feedback loops by continuously monitoring user watch history and using this data to update selection probabilities. The machine learning model processes user interactions (watching, skipping, pausing) and feeds this information back into the ranking algorithm, creating a self-improving system that adapts to individual user preferences
2Measurement precision
If videos are ordered by most recently saved or broadcast time, then the ordering is simple and deterministic, but it does not reflect actual user viewing preferences or likelihood to watch
Solution Approach 1:
The patent replaces simple mechanical ordering rules (sort by timestamp) with a machine learning-based prediction system. Instead of using deterministic time-based sorting, the system employs probabilistic models that analyze user behavior patterns to predict which videos the user is most likely to watch next, significantly improving prediction accuracy
Solution Approach 2:
The system changes the ordering parameter from static metadata (timestamp, broadcast time) to dynamic predicted selection probability. By computing and using probability scores based on user history, the system transforms the ordering criterion to better reflect actual user preferences while maintaining a manageable computational approach
3Adaptability or versatility
If all videos are presented in a single flat list, then the interface is simple, but it does not account for different types of shows or user context (e.g., current season vs. back catalog)
Solution Approach 1:
The patent segments the video list into meaningful groups based on show status and user context. Videos are divided into categories such as current season episodes, previous seasons, and shows the user is actively following versus those they are not, allowing the interface to present relevant content in an organized manner while maintaining adaptability to user needs
Data Source
AI summary
Particular embodiments provide a watch list of shows to users. The watch list is personalized for each user. Also, the watch list is dynamically organized to predict an order the user will want to watch the shows. Particular embodiments analyze historical user behavior with respect to the timing for recurring releases of the episodes for shows to determine the order of the shows in the watch list. The watch list is organized in a way that a user may select a “watch all” button where unseen episodes for the shows in the watch list are all played to the user in an order that is predicted to be the order in which the user would want to watch the shows. Providing the watch all button makes it important to predict the order of the shows accurately.