Video Ad Selection via Quality Factor Ranking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current auction systems for selecting video advertisements on video search results Web pages do not effectively consider the content and metadata of promoted video programs, leading to irrelevant ad placements, as they only focus on keywords associated with the advertisements and not the linked video content.
Innovation Solution
The proposed method involves evaluating video advertisements based on a combination of their bids, quality factors, and metadata, including unique video features, physical characteristics, and performance metrics, to rank and select relevant advertisements for display on video search results Web pages.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If auction systems only consider bids and keywords for video advertisement selection, then the selection process is simple and fast, but the relevance and effectiveness of ad placements deteriorate
Solution Approach 1:
The system pre-extracts and stores video features (duration, audio characteristics, visual characteristics) and metadata (title, description, tags) before the auction process. This preliminary preparation allows the system to quickly retrieve and evaluate relevant features during ad selection without performing complex analysis in real-time, thus maintaining speed while improving relevance through comprehensive feature consideration
Solution Approach 2:
The system transforms the traditional bid-based selection into a multi-parameter ranking system that incorporates bid amount, quality factor (derived from video features and metadata), and relevance scores. By changing from a single-parameter (bid) to multi-parameter evaluation, the system achieves both efficient processing through pre-computed features and improved ad placement relevance through comprehensive criterion consideration
2Ease of operation
If auction systems consider only bid amounts, then the selection process is straightforward, but user engagement and ad effectiveness deteriorate
Solution Approach 1:
The system segments the ad evaluation process into distinct components: bid amount evaluation, quality factor evaluation (based on video features), and relevance evaluation (based on metadata matching). This segmentation allows the system to maintain operational simplicity through modular processing while improving user engagement by considering multiple dimensions of ad quality and relevance independently
Solution Approach 2:
The system introduces a quality factor as an intermediary metric that translates complex video characteristics (duration, audio, visual features) and metadata into a unified relevance score. This intermediary quality factor serves as a mediator between the simple bid amount and the complex user engagement goals, enabling straightforward selection processes that naturally lead to more engaging ad placements
Data Source
AI summary
A method, executed on a processor, for serving a video content segment at an online resource, includes receiving a request for a video content segment; identifying one or more candidate video content segments to serve in response to the request; accessing a quality factor (QF) and a bid for each of the candidate video content segments; ranking the candidate video content segments based on a combination of each of the video content segments' QF and bid; and providing in response to the request, a set of candidate video content segments based on the ranking.