Graph-Based Translation Method for Visual Error Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current machine translation methods lack visual representation of sentence components and translation processes, making it difficult for users to understand and correct translation errors, as they only display translation results without showing the process or method used.
Innovation Solution
A method and apparatus that visually represent sentence components as nodes and edges in a graph, allowing users to interact with the graph to understand and correct translation results by designating user inputs to modify the graph structure and order, thereby providing accurate and user-confirmed translations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine translation is performed using statistical-based or neural network methods, then translation accuracy is improved, but the translation process and method information is not displayed to users
Solution Approach 1:
The patent segments the translation process into discrete visual components: sentence components are divided into nodes and edges in a graph structure, allowing users to see individual elements (words, phrases) and their relationships separately. This segmentation enables the display of translation process information while maintaining translation accuracy.
Solution Approach 2:
The patent introduces an intermediary visual representation layer (the graph with nodes and edges) between the translation engine and the user. This intermediary displays sentence components and their relationships, providing translation process information without interfering with the underlying statistical or neural network translation methods.
2Device complexity
If translation results are displayed without visual representation of sentence components, then device complexity is reduced, but user understanding and correction capability is worsened
Solution Approach 1:
The patent adds a visual dimension to translation display by creating a graph structure with nodes and edges that represents sentence components and their relationships. This dimensional transformation from text-only to visual-graph representation enables users to understand and correct translations more effectively without significantly increasing device complexity.
3Reliability
If sentence components are visually represented as nodes and edges, then user confirmation of translation accuracy is improved, but device complexity increases
Solution Approach 1:
The patent creates a universal graph representation system that can handle various sentence components (nouns, verbs, adjectives, etc.) using the same node and edge structure. This multi-functional approach allows the system to represent different language structures and relationships uniformly, improving translation confirmation accuracy while managing device complexity through standardized representation.
Data Source
AI summary
A translating method using visually represented elements, and device therefor is provided. According to an embodiment, a translation method performed by a computing device, may include acquiring data of a first sentence in a first language, the first sentence including a first morpheme of a first type and a second morpheme of a second type different from the first type, generating a first graph representing the first sentence, a first node of the first graph corresponding to the first morpheme of the first type of the first sentence, a first edge of the first graph corresponding to the second morpheme of the second type of the first sentence, and each node and each edge of the first graph being concatenated to each other so that the first sentence is completed when representations corresponding to each node and each edge of the first graph are concatenated while traversing the first graph in a first order, replacing the representations of the first language corresponding to each node and each edge of the first graph with representations of a second language, acquiring a second sentence of the second language by concatenating the representations of the second language corresponding to each node and each edge of the first graph while traversing the first graph in a second order at least in part different from the first order, and outputting the second sentence as a translation result of the first sentence into the second language.