Search Result Ranking via Watch Time Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Internet search engines struggle to effectively rank search results based on user engagement, as existing methods do not adequately account for the relevance and quality of resources by considering user watch times, leading to less relevant results being overshadowed.
Innovation Solution
A system that adjusts search result scores by incorporating user watch times, calculating watch time signals from session data to boost scores for resources users tend to watch longer and demote those watched for shorter periods, using a watch time function to recalculate and provide adjusted scores for ranking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If search results are ranked using traditional relevance algorithms, then search results can be returned quickly, but the quality and user engagement of the results deteriorate because user watch times are not considered
Solution Approach 1:
The system pre-calculates and stores watch time metrics for resources before they are needed for ranking. Watch time data is collected and processed in advance, creating a ready-to-use signal that can be quickly applied during search result ranking without adding computational overhead at query time.
Solution Approach 2:
The patent introduces watch time as an intermediary metric that bridges user behavior and search result ranking. Instead of directly analyzing complex user interaction patterns during search, the system uses pre-computed watch time signals as a mediator to inform ranking decisions, simplifying the connection between user engagement and result quality.
2Reliability
If search results are adjusted to promote resources with longer watch times, then resource quality and relevance improve, but the computational resources and time required for scoring increase
Solution Approach 1:
Watch time metrics are calculated and stored in advance for all resources in the index. This pre-computation allows the ranking system to quickly retrieve and apply watch time signals during search operations without performing complex calculations at query time, significantly reducing score calculation time.
Solution Approach 2:
The system transforms raw watch time data into normalized scoring parameters that can be efficiently applied during ranking. By converting watch time into standardized signals with defined weightings, the system enables fast mathematical operations during score calculation while maintaining reliable relevance assessment.
3Measurement precision
If watch time data is collected and processed for all search resources, then ranking accuracy improves, but the data processing complexity and storage requirements increase
Solution Approach 1:
The system extracts only the essential watch time metric from complex user interaction data, discarding unnecessary details about individual user behaviors. By focusing on the aggregate watch time signal rather than processing every user action, the system achieves precise engagement measurement with reduced processing complexity.
Solution Approach 2:
The patent applies different levels of watch time analysis to different resources based on their importance and characteristics. High-value resources receive more detailed watch time analysis while less critical resources use simplified metrics, optimizing the balance between measurement precision and processing complexity across the entire system.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for ranking search results. One of the methods includes identifying one or more sessions for a query and associating watch times of the respective resources watched in the sessions with the query. One or more watch time signals are calculated for a first resource and the query based on the watch times associated with the query. A first search result responsive to the query is obtained, wherein the first search result identifies the first resource and has an associated score S. A new score S′ is calculated based on a least S and a watch time function, the watch time function being a function of the one or more watch time signals. The new score S′ is provided to a process for ranking search results including the first search result.