Video Recommendation Keyword Extraction and Ranking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video recommendation systems on online video platforms often fail to provide relevant recommendations due to similarities in video content, leading to unsuitable thumbnails or lengthy videos, which do not align with users' browsing preferences within a limited time frame.
Innovation Solution
A video recommendation system that extracts keywords from video titles, compares them with target keywords, and generates similarity scores to rank candidate videos, selecting recommendations based on these scores, while considering secondary ranking criteria such as source, duration, and viewer engagement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If video recommendations are based on similarity of video content, then content-related videos are recommended, but the recommendations may have unsuitable thumbnails or be too long to watch
Solution Approach 1:
The patent segments the video recommendation problem into multiple independent ranking factors: content similarity, thumbnail quality, duration, and viewer engagement metrics. Each factor is evaluated separately and combined to produce a comprehensive ranking, allowing the system to balance content relevance with user-friendly attributes.
Solution Approach 2:
The patent changes the ranking parameters from solely content-based similarity to a multi-parameter system that includes metadata attributes such as duration thresholds, thumbnail quality scores, and engagement metrics. This allows the system to filter out videos that are too long or have poor thumbnails while maintaining content relevance.
2Reliability
If video recommendations are based on content similarity, then related videos are identified, but users cannot browse many videos within a limited time period
Solution Approach 1:
The patent performs preliminary filtering and ranking of candidate videos based on multiple criteria before presenting them to users. Videos are pre-evaluated for duration, thumbnail quality, and engagement metrics, so that only videos meeting certain thresholds are recommended, saving users time during actual browsing.
Solution Approach 2:
The patent applies partial action by selectively applying different ranking weights to different video attributes based on user context. For example, when a user is in quick-browsing mode, duration and thumbnail quality are weighted more heavily, while content similarity is weighted less, allowing faster presentation of suitable videos.
3Reliability
If multiple ranking criteria are used to improve recommendation quality, then more comprehensive video evaluation is achieved, but system complexity increases
Solution Approach 1:
The patent implements a universal ranking framework that can evaluate videos across multiple dimensions using a single integrated scoring system. The same ranking mechanism handles content similarity, duration filtering, thumbnail quality assessment, and engagement metric evaluation, reducing the need for separate complex systems for each criterion.
Data Source
AI summary
A solution is provided to generate video recommendations in a video sharing environment. A video recommendation system selects a video as a target video and extracts target keywords from the title of the identified target video or the title of a non-video trending news item. The system receives multiple candidate videos. For each candidate video, the system extracts keywords from the title of the candidate video and compares the extracted keywords with the target words. Based on the comparison, the system generates a similarity score for the candidate video. The system ranks the candidate videos based their associated similarity scores and selects a candidate video having the highest similarity score as the video recommendation for the target video.