Target-Video Co-Watched Clusters for Recommendation Diversity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Video hosting websites lack effective methods to recommend diverse and relevant videos to users based on their viewing patterns and search queries, often failing to provide personalized and contextually accurate suggestions.
Innovation Solution
The system clusters videos watched by a user within a specific time window based on associated keywords, generating target-video specific clusters that can be used to select and present related videos, advertisements, and adjust search result rankings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If video hosting websites display only basic video information and simple related video links, then the system complexity remains low, but the recommendation diversity and personalization are insufficient
Solution Approach 1:
The patent segments co-watched videos into multiple clusters based on different search queries and user interaction patterns. Each cluster represents a distinct theme or category of related videos, allowing the system to provide diverse recommendations without requiring a single complex recommendation engine. This segmentation enables the system to handle complexity in a modular fashion while improving recommendation versatility.
Solution Approach 2:
The patent introduces a new dimension of organization by clustering videos according to search queries and user behavior patterns rather than traditional categorical classification. This adds a behavioral dimension to video recommendations, enabling personalization based on how users actually search and watch videos, thereby improving adaptability without proportionally increasing system complexity.
2Adaptability or versatility
If the system collects and processes detailed user interaction data for personalized recommendations, then recommendation personalization improves, but data processing complexity and computational resources increase
Solution Approach 1:
The patent extracts specific, high-value signals from user interaction data—namely, which videos users watch after searching for particular queries. Rather than processing all possible user behavior data, the system focuses on extracting co-watched video patterns associated with search queries, thereby achieving personalization with reduced data processing complexity.
Solution Approach 2:
The system performs preliminary clustering of videos by search query and user interaction patterns during off-peak times or in advance, storing these pre-computed clusters for rapid retrieval during user sessions. This preliminary action reduces real-time processing complexity while maintaining high personalization capabilities.
3Measurement precision
If the system generates and stores multiple clusters of co-watched videos for each target video, then recommendation relevance improves, but storage requirements and data management complexity increase
Solution Approach 1:
The patent creates video clusters that serve multiple functions: they enable personalized recommendations, support search result augmentation, and provide the basis for generating related video suggestions. This multi-functionality allows the same clustered data structure to improve recommendation relevance across different system components without proportionally increasing storage requirements.
4Productivity
If the system presents augmented search results with additional co-watched videos, then user engagement improves, but information overload and user confusion may increase
Solution Approach 1:
The patent segments augmented search results into distinct clusters based on search queries and user behavior patterns. Each cluster represents a coherent theme of related videos, allowing users to explore additional content in an organized manner rather than receiving a undifferentiated list, thereby reducing information overload while maintaining engagement.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for clustering videos and using video clusters to present content. One method includes generating target video-specific clusters of co-watched videos for target videos, according to keywords for each target video, and storing data associating the target video with its clusters. The clusters can then be used in various applications, including identifying related videos, augmenting video search results, re-ordering video search results, and identifying content for a target video.