Extended Search Method for Personalized Long-Tail Queries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing personalized search methods struggle to effectively meet long-tail user requirements and are resource-intensive, particularly in scenarios like picture search where content description data is sparse, leading to limitations in search result diversity and accuracy.
Innovation Solution
An extended search method that establishes interest and extended term models based on user behavior logs, allowing for personalized search result extension by determining relevant search terms from interest and extended term models, thereby improving search result relevance and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If natural search results are rearranged based on user interest to improve personalization, then search result relevance to user interest is improved, but search result diversity and coverage of long-tail requirements deteriorate
Solution Approach 1:
The patent segments search results into two categories: natural search results (top-n) and extended search results (from long-tail). By dividing the search space and applying different processing methods to each segment, the system maintains relevance for popular queries while preserving diversity for long-tail requirements, resolving the contradiction between personalization and diversity
Solution Approach 2:
The patent adds a new dimension to search by introducing extended search results from long-tail data that complement the traditional natural search results. This dimensional expansion allows the system to simultaneously satisfy both popular group requirements and individual long-tail needs, achieving both relevance and diversity
2Speed
If only few front results are intercepted for rearrangement to ensure search response time, then search response speed is improved, but coverage of user requirements deteriorates
Solution Approach 1:
The patent pre-processes and stores long-tail search results in advance, organizing them into structured data that can be quickly retrieved during search operations. This preliminary preparation enables fast response times while maintaining comprehensive coverage of user requirements, as the extended results are ready for immediate delivery without real-time processing overhead
3Measurement precision
If abundant data and expensive calculation storage costs are incurred to extract characteristics for personalized arrangement, then personalization accuracy is improved, but calculation and storage costs increase
Solution Approach 1:
The patent extracts only the essential characteristics needed for personalization from user behavior data, rather than processing all available data. By selectively extracting relevant features (such as user interests and preferences) and discarding redundant information, the system achieves accurate personalization with reduced calculation and storage costs
Solution Approach 2:
The patent uses lightweight, simplified data structures and models for personalization that require minimal storage resources and computation power. Instead of maintaining complex, resource-intensive models, the system employs efficient algorithms that achieve acceptable personalization accuracy with significantly lower costs
4Measurement precision
If manual or machine learning methods are used to establish subject model for user interest description, then interest similarity calculation quality is improved, but construction and update costs increase
Solution Approach 1:
The patent implements automated subject model construction and updating using machine learning algorithms that learn from user behavior data without requiring manual intervention. The model self-updates as new data becomes available, eliminating the need for expensive manual maintenance while maintaining high calculation quality through continuous learning from real user interactions
Data Source
AI summary
An extended search method and apparatus is provided. An interest term model of each user is established. An extended term model of each fourth search sequence is established. A corresponding extended search term is determined based on a current search sequence of a current user and based on an interest term model of the current user as well as an extended term model of the current search sequence; and a corresponding search result is provided for the current user based on the current search sequence and the extended search term. The present invention can implement simpler and more efficient personalized searches, and is advantageous in terms of meeting long-tail requirements of users (that is, adding personalized result recalling), reducing calculation and storage costs, universality, practicability, and the like.