Amendment Source Positioning via Semantic Parsing and Fuzzy Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing amendment source-positioning methods in speech input applications are inflexible and inefficient, limited by template matching, which struggles to accurately position correction sources in varied text contexts.
Innovation Solution
A method and apparatus that utilize semantic parsing to identify and position amendment sources through multiple fuzzy positioning techniques, including synonymy, similarity, and phonetic notation transformations, to enhance flexibility and accuracy in error correction within speech input applications.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If template matching and positioning method is used, then the amendment source can be positioned according to preset templates, but the method is rigid and inflexible with lower positioning efficiency
Solution Approach 1:
The patent transforms the rigid template matching approach into a flexible parameter-based positioning system. Instead of fixed templates, the system uses semantic parsing to extract parameters (amendment type, target object, position information) from speech instructions, allowing dynamic adaptation to various correction scenarios while maintaining efficient positioning through structured parameter extraction and matching.
2Measurement precision
If strict template matching is used, then positioning accuracy can be maintained for preset patterns, but it cannot handle parsing errors or pronunciation variations
Solution Approach 1:
The patent implements error tolerance mechanisms beforehand in the positioning process. The system pre-prepares for potential parsing errors and pronunciation variations by incorporating fuzzy matching algorithms and multiple candidate generation. When exact matches fail, the system can fallback to similarity-based matching, ensuring continuous operation without complete failure.
Solution Approach 2:
The system incorporates feedback mechanisms where positioning results are evaluated and can trigger refinement processes. If initial positioning based on semantic parsing fails or produces low-confidence results, the system uses feedback to activate alternative positioning strategies, such as fuzzy matching or phonetic similarity search, thereby maintaining accuracy despite input variations.
Data Source
AI summary
An amendment source-positioning method and apparatus, a computer device and a readable medium. The method includes: obtaining a first target word identifying an amendment source and defining parameters of the amendment source, from semantic parsing information of a user-input speech error correction instruction; positioning the amendment source from a to-be-corrected text according to the first target word and the defining parameters. As compared with the template matching and positioning scheme employed in the prior art, the technical solution of the present disclosure can support a speech error correction instruction in any form, and exhibits a more flexible amendment source-positioning manner, thereby effectively improving the amendment source-positioning efficiency.