Semantic Parsing System Using Coarse-Grained Domain Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing semantic parsing systems face challenges in data management, low parsing efficiency, and accuracy due to variable application scenarios and the complexity of user statements across multiple vertical domains.
Innovation Solution
The method involves obtaining a first recognition result for a target statement, determining a target vertical domain based on the intention recognition result, converting the entity recognition result into a second entity recognition result within the target domain, and parsing the intention of the statement accordingly. This approach uses coarse-grained recognition at an upper level to improve parsing efficiency and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If semantic parsing is performed across multiple vertical domains simultaneously, then the system can handle diverse application scenarios, but data management complexity increases and parsing efficiency decreases
Solution Approach 1:
The patent segments the semantic parsing system into domain-specific modules, where each vertical domain (e.g., healthcare, finance, legal) has its own dedicated parsing component. This segmentation allows the system to handle diverse scenarios by routing queries to appropriate domain modules, improving parsing efficiency within each domain while maintaining overall versatility.
Solution Approach 2:
The patent introduces an intermediary layer (domain adapter or routing mechanism) that sits between the general query interface and domain-specific parsers. This intermediary determines which vertical domain a query belongs to and directs it to the appropriate parser, enabling efficient domain-specific processing while maintaining the ability to handle multiple domains.
2Adaptability or versatility
If semantic parsing is performed across multiple vertical domains simultaneously, then the system can handle diverse application scenarios, but data management complexity increases
Solution Approach 1:
The patent implements a universal data management framework that can handle multiple vertical domains through standardized interfaces and common data structures. This universal layer provides domain-agnostic functions for data storage, retrieval, and management, reducing the complexity that would otherwise arise from managing separate systems for each domain.
Solution Approach 2:
The patent merges the data management functionality across all vertical domains into a unified system with standardized schemas and interfaces. By combining domain-specific data requirements into a common framework with extensible capabilities, the system reduces management complexity while maintaining support for diverse domains.
3Measurement precision
If fine-grained entity recognition is performed for all domains, then parsing accuracy is improved, but processing time increases
Solution Approach 1:
The patent implements dynamic entity recognition that adapts the level of granularity based on the detected domain and query context. For example, healthcare queries trigger detailed medical entity recognition, while general queries use coarser recognition. This dynamic adjustment maintains high accuracy for domain-specific queries while reducing processing time for less demanding queries.
Solution Approach 2:
The patent applies different levels of entity recognition granularity to different domains and query types. High-precision fine-grained recognition is applied only where necessary (specific vertical domains with well-defined schemas), while coarser recognition is used for other cases, optimizing the balance between accuracy and processing time.
Data Source
AI summary
Methods, electronic device, and non-transitory computer-readable storage mediums are provided for semantic parsing. The equipment may obtain a first recognition result of a target statement. The first recognition result may include a first intention recognition result and a first entity recognition result. The first entity recognition result may correspond to a plurality of vertical domains. The equipment may also determine one of the plurality of vertical domains corresponding to the first entity recognition result as a target vertical domain corresponding to the target statement according to the first intention recognition result. The equipment may further convert the first entity recognition result into a second entity recognition result in the target vertical domain. The equipment may also parse an intention of the target statement according to the second entity recognition result.