GBRank Model for Query Result Bottom Retrieval
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional ranking methods in online search engines, such as bucket sort, are limited by few base relevance characteristics, poor scalability, and inability to effectively differentiate resources with multiple characteristics, leading to suboptimal query result relevance.
Innovation Solution
The use of a Gradient Boosting Rank (GBRank) model to acquire and rank query resources based on comparative scoring characteristics, including base relevance, quality control, and click characteristics, which learns relationships from training data using machine learning, allowing for automatic updates and improved differentiation of resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If traditional bucket sort method is used for ranking, then the system is simple to implement, but the retrieving effect is poor and scalability is low
Solution Approach 1:
The patent replaces the mechanical bucket sort algorithm with a machine learning-based GBRank model. This substitution enables the system to automatically learn optimal ranking strategies from data, improving scalability and adapting to different query types without requiring manual redesign of the ranking logic.
Solution Approach 2:
The patent changes the fundamental parameters of the ranking system by introducing multiple base relevance characteristics (text matching, URL matching, domain matching, etc.) and using a machine learning model to dynamically weight them. This allows the system to adapt to different scenarios by adjusting parameter weights based on learned patterns rather than fixed manual configurations.
2Measurement precision
If manually analyzing association degree between base relevance characteristics and resource relevance is performed, then the ranking can be optimized, but the process is time-consuming and not convenient when adding or reducing characteristics
Solution Approach 1:
The GBRank model enables the system to automatically analyze and determine the association degrees between base relevance characteristics and resource relevance through machine learning. The model self-adjusts the weights of different characteristics based on training data, eliminating the need for manual analysis and making the system automatically adaptable when new characteristics are added.
Solution Approach 2:
The patent implements a feedback mechanism where the GBRank model continuously learns from query results and user interactions, automatically adjusting the weighting of base relevance characteristics. This feedback loop replaces manual analysis by using actual system performance data to optimize ranking precision dynamically.
3Reliability
If more base relevance characteristics are adopted in bucket sort model, then the comprehensive relevance improves, but the influence of characteristics ranked behind decreases and cannot make correction
Solution Approach 1:
The patent introduces dynamic weighting of base relevance characteristics through the GBRank model. Instead of fixed hierarchical weighting in bucket sort, the machine learning model dynamically adjusts the importance of each characteristic based on the specific query and resource context, allowing all characteristics to contribute meaningfully regardless of their position in any fixed hierarchy.
Solution Approach 2:
The patent combines multiple base relevance characteristics (text matching, URL matching, domain matching, click characteristics, quality control characteristics) into a composite scoring system. The GBRank model integrates these diverse characteristics with learned weights, creating a composite relevance score that leverages the strengths of each characteristic without being constrained by hierarchical limitations.
4Speed
If simple base relevance characteristics are used for ranking, then the ranking process is fast, but the retrieving effect in long queries is poor
Solution Approach 1:
The GBRank model serves multiple functions simultaneously: it processes various types of base relevance characteristics (text, URL, domain matching), handles different query types (including long queries), and adapts to different scenarios. This multi-functionality allows the system to maintain fast processing while improving retrieving effect across diverse query complexities.
Solution Approach 2:
The patent adds new dimensions to the ranking process by incorporating click characteristics and quality control characteristics alongside traditional text matching. The GBRank model processes these additional dimensions to improve retrieving effect for complex queries while maintaining efficient processing through automated machine learning-based scoring.
Data Source
AI summary
Embodiments of the present invention disclose a query result bottom retrieval method and apparatus. The query result bottom retrieval method includes: acquiring, from a resource library and based on a user input target query, query resources associated with the user input target query; acquiring comparative scoring characteristics of each of the query resources, the comparative scoring characteristics including base relevance characteristics; inputting the comparative scoring characteristics of each of the query resources into a pre-trained rank model, to acquire relevance scoring values corresponding to the query resources and output by the rank model, the rank model being a GBRank model; and ranking the query resources based on the relevance scoring values, and selecting, according to the ranking result, a set number of target resources as a bottom retrieval result corresponding to the target query. The technical solution of the present invention, by using a GBRank rank model, optimizes a traditional query result bottom retrieval method, and enhances relevance between a retrieved target resource and a target query.