AI Search Query Understanding via Semantic Pattern Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional search engines lack deep understanding of user queries, leading to low accuracy in retrieving relevant information, as they rely on keyword matching and user click history, failing to account for semantic meaning and user intent.
Innovation Solution
A search method based on artificial intelligence that performs word segmentation, structured analysis to generate a semantic pattern, and knowledge-based analysis to determine an understanding result, enabling deeper query understanding and improved relevance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional keyword matching and user click history are used for search, then the search system is simple and fast, but the accuracy of retrieving relevant information is low
Solution Approach 1:
The query understanding process is segmented into multiple analysis dimensions: semantic pattern analysis, knowledge-based analysis, and user intent analysis. Each dimension processes specific aspects of the query independently, then results are integrated to form a comprehensive understanding result, improving accuracy without creating a monolithic complex system
Solution Approach 2:
The patent adds new analysis dimensions beyond traditional keyword matching. It introduces semantic patterns (syntactic structure, semantic relationships), knowledge-based analysis (entity recognition, relationship extraction), and intent recognition (information need, query type), transforming the search from 1D keyword matching to multi-dimensional query understanding
2Loss of information
If keyword matching is used for query analysis, then the processing speed is fast, but the understanding of user intent and semantic meaning is insufficient
Solution Approach 1:
The system performs preliminary word segmentation and semantic pattern generation on the query before the actual search. By pre-processing the query into structured semantic patterns and extracting key entities, the system prepares understanding frameworks in advance that guide the search process, reducing information loss while managing processing time efficiently
Solution Approach 2:
The patent introduces semantic patterns as an intermediary representation between raw keywords and search results. These patterns act as mediators that capture semantic meaning and user intent, bridging the gap between simple keyword matching and comprehensive understanding, allowing the system to retain semantic information without excessive processing overhead
3Measurement precision
If deep semantic analysis and knowledge-based analysis are performed on queries, then the understanding accuracy is improved, but the computational complexity increases
Solution Approach 1:
The complex analysis process is divided into separate modules: semantic pattern analysis module, knowledge-based analysis module, and intent recognition module. Each module handles specific analytical tasks independently, allowing for optimized processing of each dimension and easier management of overall system complexity
Solution Approach 2:
The patent creates a universal query understanding framework that handles multiple types of queries through the same semantic pattern and knowledge-based analysis mechanisms. This multi-functional approach improves understanding accuracy across diverse query types while avoiding the need for separate complex systems for each query category
Data Source
AI summary
The present disclosure provides a search method and device based on artificial intelligence and an electronic device. The search method based on artificial intelligence includes: obtaining a query; performing a word segmentation on the query to obtain a term sequence containing a plurality of terms; performing a structured analysis on the term sequence to generate a semantic pattern; performing a knowledge-based analysis on the term sequence based on the semantic pattern to generate a semantic analysis result; determining an understanding result corresponding to the query based on the semantic pattern and the semantic analysis result; and performing a search based on the understanding result corresponding to the query.