Video Clip Popularity Scoring for Live Stream Engagement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Viewers face difficulty in determining the most desirable video clips among a large pool of stream clips, as existing video streaming services lack efficient methods to measure and showcase clip popularity, leading to limited user engagement and exposure for streamers and game publishers.
Innovation Solution
A video clip popularity measuring system that computes popularity scores based on viewer actions, merges data from overlapping clips, and selects clips for viewers based on desired characteristics, allowing streamers to create streamer-generated video items that are intelligently timed and targeted to live game sessions, enhancing interaction and engagement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If a large pool of stream clips is provided to viewers, then the quantity of content available increases, but viewers face difficulty in determining the most desirable video clips
Solution Approach 1:
The system implements feedback mechanisms by tracking viewer actions (plays, shares, comments, likes) on video clips and using this feedback to compute popularity scores. These scores are then displayed to help other viewers identify desirable clips, solving the information asymmetry problem in large clip pools.
Solution Approach 2:
The system uses visual indicators (such as popularity scores, badges, or highlighting) to differentiate desirable clips from others in the large pool, making it easy for viewers to identify high-quality content without manually evaluating each clip.
2Measurement precision
If popularity measures are computed for all video clips, then clip popularity can be determined, but computational resources and time are consumed
Solution Approach 1:
The system computes popularity measures for clips based on actual viewer engagement rather than evaluating all possible clips uniformly. It focuses computational resources on clips that have been viewed or interacted with, rather than computing scores for the entire pool of clips.
Solution Approach 2:
The system pre-computes and stores popularity scores as clips accumulate viewer actions, rather than computing them on-demand. This allows popularity information to be readily available when viewers browse clips, eliminating computation delays during user interactions.
3Productivity
If streamer-generated video items are created and displayed, then user engagement increases, but the system complexity increases
Solution Approach 1:
The system uses a unified popularity measurement framework that serves multiple functions: ranking clips for display, selecting clips for streamer-generated video items, and providing feedback to viewers. This multi-functional approach increases engagement without proportionally increasing system complexity.
Solution Approach 2:
The system automatically generates video items using popularity scores and viewer action data without requiring manual curation. The automated processes handle clip selection, video item creation, and display timing, reducing the operational complexity despite increased functionality.
Data Source
AI summary
Viewers of a video stream may generate video clips including different portions of the video stream. Popularity measures may be computed for the video clips, for example based on tracked actions associated with the video clips, such as frequency of playing and sharing of the video clips. The popularity measures may be used to select and provide video clips to viewers. Video items may be generated by streamers, such as may include selected portions of video of streamers playing a game. A video item generated by a particular streamer may be displayed to viewers only during times when that particular streamer is participating in an active game session. Viewers of the video item may provide input that allows viewers to receive a live stream of the streamer's active game session that it is being played simultaneously with the display of the video item.