Query Relevance Scoring via Category-Specific Selection Frequencies
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current internet search engines face challenges in accurately determining user intent and relevance of search results, as they rely on general assumptions about user preferences and do not effectively utilize specific user behaviors and interests to adjust rankings.
Innovation Solution
The system processes prior queries and user activities to generate adjusted scores for search results based on category-specific and general selection frequencies, incorporating demographic and location information to refine search result rankings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If general assumptions about user preferences are used to rank search results, then the search engine can process queries efficiently, but the relevance and accuracy of search results deteriorates
Solution Approach 1:
The patent segments users into different categories based on their search behavior patterns, demographics, and location. By dividing the user base into distinct segments (e.g., mobile users, desktop users, location-specific users), the system can apply customized ranking strategies to each segment rather than using a single general approach, thereby improving relevance while maintaining efficiency through automated segmentation.
Solution Approach 2:
The patent dynamically changes ranking parameters based on user characteristics, device type, location, and search history. Instead of using fixed ranking parameters, the system adjusts parameters such as recency weighting, location relevance, and user preference factors according to the specific user context, improving measurement precision without significantly impacting processing efficiency.
2Measurement precision
If user-specific clues such as device type and location are incorporated into search results, then search result relevance improves, but system complexity increases
Solution Approach 1:
The patent implements a universal ranking framework that handles multiple user characteristics (device type, location, search history, demographics) through a single multi-functional system. Rather than creating separate systems for each user attribute, the ranking algorithm integrates all these factors into one unified process, improving relevance while avoiding the complexity of multiple independent systems.
Solution Approach 2:
The system automatically collects and processes user-specific information (device type, location, search patterns) without requiring manual input or complex configuration. The ranking algorithm self-adjusts based on the data it receives, reducing the need for manual system management and lowering operational complexity while maintaining high relevance.
3Measurement precision
If search results are adjusted based on category-specific selection frequencies, then relevance for specific user categories improves, but the complexity of score adjustment increases
Solution Approach 1:
The patent pre-calculates and stores selection frequency statistics for different user categories and search result categories during off-peak times. By performing this computation in advance rather than in real-time, the system reduces the complexity of score adjustment during actual search operations, as the pre-computed statistics can be directly applied without complex real-time calculations.
Solution Approach 2:
The patent introduces selection frequency statistics as an intermediary layer between user queries and search result ranking. Instead of directly analyzing complex user behavior patterns in real-time, the system uses pre-computed selection frequencies as a mediator to adjust rankings, simplifying the scoring process while maintaining category-specific relevance.
Data Source
AI summary
A computer-implemented method for processing query information includes receiving prior queries followed by a current query, the prior and current queries being received within an activity period an originating with a search requester. The method also includes receiving a plurality of search results based on the current query. Each search result identifying a search result document, each respective search result document being associated with a query specific score indicating a relevance of the document to the current query. The method also includes determining a first category based, at least in part, on the prior queries. The method also includes identifying a plurality of prior activity periods of other search requesters, each prior activity period containing a prior activity query where the prior activity query matches the current query, and where the prior activity period indicates the same first category.