Search Engine Relevancy Scoring via Predicted Hybrid Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Collaboration platforms, such as gaming platforms, face issues with providing relevant search results due to inappropriate or insufficient historical data, leading to user frustration and reduced engagement, as existing search engines often return irrelevant content items with little or no historical interaction data.
Innovation Solution
The proposed solution involves scoring search terms based on historical data, including the frequency of content items being returned and selected by users, to determine relevant search results, allowing content items with little historical data to surface, thereby improving user experience and search efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If search engines use historical data to rank content items, then search result relevancy is improved, but content items with little or no historical data are excluded from results
Solution Approach 1:
The patent changes the parameter used for ranking from purely historical interaction data to a hybrid model that incorporates predicted relevancy scores. This allows content items with little historical data to be included by using predicted scores generated from alternative data sources or models, thus resolving the contradiction between requiring historical data for accuracy and needing to include new content items.
Solution Approach 2:
The patent introduces predicted relevancy scores as an intermediary between historical data and search ranking. This intermediary component enables the system to rank content items even when historical data is insufficient, by using predictions derived from other data sources or models, thereby maintaining search result quality while including new content items.
2Measurement precision
If search engines rely on historical interaction data, then search accuracy is improved, but user engagement is reduced due to irrelevant results for new content
Solution Approach 1:
The patent changes the ranking parameter to incorporate predicted relevancy scores alongside historical data. This hybrid approach maintains search accuracy for established content while improving user engagement by ensuring new content items can surface in search results, thereby preventing user frustration and reducing churn.
Solution Approach 2:
The patent implements a feedback mechanism where predicted relevancy scores are continuously refined based on user interactions. This allows the system to learn from user behavior patterns and improve predictions, thereby maintaining high search accuracy while increasing user engagement through more relevant results for both established and new content.
3Measurement precision
If search engines use comprehensive historical data analysis, then search result quality is improved, but computational resources and storage requirements increase
Solution Approach 1:
The patent extracts and pre-computes predicted relevancy scores as a separate component from full historical data analysis. This extraction allows the system to use lightweight predicted scores for initial ranking decisions, reducing the need to process and store comprehensive historical data for every search query, thus improving search result quality while reducing computational and storage requirements.
Solution Approach 2:
The patent segments the search ranking process into multiple stages: first using predicted relevancy scores for initial filtering and ranking, then applying more comprehensive historical data analysis only when necessary. This segmentation reduces the overall computational and storage resources required while maintaining high search result quality through the use of predicted scores as a first-pass ranking mechanism.
Data Source
AI summary
A method includes identifying a search term and obtaining historical data indicative of a number of times a game was selected from search results corresponding to search queries using the search term and indicative of a number of times the selected game was played by one or more users for at least a threshold amount of time. The method further includes generating a score based on the historical data. The score is associated with the search term and the selected game.