Query Processing Method for Intent Determination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for natural language understanding in artificial intelligence face challenges in extracting text due to the diversity and complexity of semantic and sentence structures, primarily focusing on keyword extraction rather than accurately determining user intent.
Innovation Solution
A query processing method that determines a target candidate entity set based on query feedback and a preset knowledge base, using a dual-channel model to evaluate and categorize entities, thereby accurately selecting the search intent entity and reducing search costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If keyword extraction methods are used for text analysis, then the processing speed is maintained, but the accuracy of user intent determination deteriorates
Solution Approach 1:
The query processing method is segmented into distinct stages: initial keyword extraction to generate candidate entities, followed by evaluation and categorization stages. This segmentation allows the system to maintain processing efficiency at the extraction stage while improving accuracy through subsequent specialized evaluation steps, resolving the contradiction between speed and accuracy.
Solution Approach 2:
Candidate entities serve as an intermediary between raw keyword extraction and final user intent determination. The system extracts keywords, generates candidate entities from them, evaluates these candidates, and then determines user intent based on the evaluation results. This intermediary layer enables both efficient processing and accurate determination by breaking down the complex task into manageable steps.
2Reliability
If comprehensive entity evaluation is performed to improve search accuracy, then the recall rate improves, but the search cost increases
Solution Approach 1:
The system performs preliminary evaluation of candidate entities before final selection, assessing their relevance to the query in advance. This preliminary action filters out irrelevant candidates early in the process, ensuring that only promising candidates undergo more intensive evaluation, thereby improving recall rate while controlling processing time.
Solution Approach 2:
The evaluation process applies different levels of scrutiny to different candidate entities based on their local characteristics. High-priority candidates receive more comprehensive evaluation while lower-priority ones are processed more quickly. This localized quality approach ensures thorough evaluation where needed while maintaining overall processing efficiency.
Data Source
AI summary
The present disclosure provides a query processing method. The method includes: determining a query feedback based on a received query request including a character string; determining a target candidate entity set based on the query feedback, the character string and a preset knowledge base; evaluating each target candidate entity in the target candidate entity set to obtain evaluation results; and determining a category of each target candidate entity based on the evaluation result of the target candidate entity.