Content Ranking Using Last-Viewed Loss Scores
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Determining the relevance of content elements for users is challenging, as existing methods struggle to accurately predict user interest, leading to inefficient content recommendation.
Innovation Solution
A ranking system that analyzes user interactions with content elements, assigning win and loss scores, and uses machine learning to predict relevance scores based on these interactions, thereby determining optimal rank positions for content elements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional content recommendation methods are used, then the system is simple to implement, but the accuracy of predicting user interest is low
Solution Approach 1:
The patent segments the content recommendation problem into distinct components: win score calculation (for interacted content) and loss score calculation (for non-interacted content). This segmentation allows the system to handle different types of user feedback separately, improving prediction accuracy while maintaining manageable system complexity through modular processing of user interactions
Solution Approach 2:
The patent applies preliminary action by pre-calculating win scores and loss scores for content elements before final ranking. The system prepares training data with these pre-computed scores, which are then used to train the ranking model in advance, enabling accurate real-time recommendations without complex computations during user interactions
2Reliability
If content elements are ranked based on limited user interaction data, then the system operates quickly, but the relevance of recommended content is poor
Solution Approach 1:
The patent performs preliminary training of the ranking model using historical user interaction data before deployment. Win scores and loss scores are calculated in advance from past interactions, and the model is trained offline. This preliminary action ensures high reliability of recommendations while keeping real-time operation fast, as the heavy processing occurs beforehand
Solution Approach 2:
The patent implements feedback mechanisms where user interactions (clicks, views, skips) continuously provide signals that update win and loss scores. This feedback loop allows the system to learn from actual user behavior, improving content relevance over time while processing updates efficiently through incremental learning from interaction patterns
3Measurement precision
If the system analyzes all user interactions in detail, then the prediction accuracy improves, but the processing complexity and time increase
Solution Approach 1:
The patent segments user interactions into two distinct categories: positive interactions (wins) where users engage with content, and negative interactions (losses) where users skip or ignore content. This segmentation simplifies the analysis complexity by creating clear, discrete signal types while maintaining precise prediction through the combined use of win scores and loss scores in the ranking model
Solution Approach 2:
The patent transforms raw user interaction data into standardized numerical parameters (win scores and loss scores) that can be directly processed by the ranking model. This parameter transformation simplifies the complexity of analyzing diverse interaction types while preserving prediction precision through normalized, quantifiable metrics that capture the essence of user preference
Data Source
AI summary
A method and system for determining loss scores for content elements displayed on an interface are provided. The method comprises: displaying, to a user, the interface including a plurality of content elements; receiving an indication of user interactions with the interface, wherein the user interactions include a shutdown or refresh of the interface; determining, a last-viewed content element of the plurality of content elements, wherein the last-viewed content element was a last element interacted with by the user prior to the shutdown or the refresh of the interface; determining one or more content elements, of the plurality of content elements, that were displayed to the user below the last-viewed content element; and determining, for each of the one or more content elements, a loss score corresponding to the respective content element based at least in part on a position of the respective content element below the last-viewed content element.