Video Recommendation via Playlist Co-occurrence Statistics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video recommendation systems in online video sharing platforms often fail to recommend videos that do not receive enough co-watches, limiting the visibility of user-uploaded content that may be of interest to only a limited number of viewers.
Innovation Solution
A video recommendation system that computes video co-occurrence data to select and rank videos based on their frequency of appearance together on playlists, providing recommendations for underwatched videos by analyzing metadata and co-occurrence statistics to increase their visibility.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If co-visitation based video recommendation is used to select videos being watched together frequently, then the recommendation accuracy for popular videos is improved, but the visibility of underwatched user-uploaded videos deteriorates
Solution Approach 1:
The patent changes the recommendation parameter from co-visitation frequency to video co-occurrence statistics in playlists. Instead of measuring how often videos are watched together, it measures how often videos appear together in user-created playlists, which better captures intentional pairing and thematic relationships, thereby improving visibility for niche videos while maintaining recommendation quality
Solution Approach 2:
The patent introduces video playlists as an intermediary structure that connects videos. By analyzing co-occurrence within this intermediary framework, the system can discover relationships between videos that share thematic or contextual connections even if they haven't been watched together frequently, thus promoting underwatched videos that belong to specific niches or categories
2Productivity
If video recommendation is based on co-watches, then the system can identify related videos efficiently, but videos with limited viewer interest are excluded from recommendations
Solution Approach 1:
The patent performs preliminary analysis of video co-occurrence in playlists to pre-establish recommendation relationships. By analyzing playlist compositions in advance and building recommendation indices based on co-occurrence patterns, the system prepares recommendation data beforehand, enabling efficient retrieval while capturing niche video relationships that co-watch metrics would miss
Solution Approach 2:
The patent transitions from a single-dimension co-watch metric to a multi-dimensional approach by incorporating playlist co-occurrence data. This adds a new dimension of analysis that captures intentional video pairings and thematic groupings, allowing the system to discover and recommend niche videos through their contextual relationships in playlists rather than relying solely on viewing behavior frequency
Data Source
AI summary
A system and method provides video recommendations for a target video in a video sharing environment. The system selects one or more videos that are on one or more video playlists together with the target video. The video co-occurrence data of the target video associates the target video and another video on one or more same video playlists and frequency of the target video and another video on the video playlists is computed. Based on the video co-occurrence data of the target video, one or more co-occurrence videos are selected and ranked based on the video co-occurrence data of the target video. The system selects one or more videos from the co-occurrence videos as video recommendations for the target video.