Text Sequence Generation Using Structured Feature Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automatic text conversion technologies, such as text-to-text and table-to-text conversion, fail to accurately and fluently edit text content, leading to factual errors and disordered vocabulary due to their inability to correct and refine text sequences effectively.
Innovation Solution
A text sequence generating method that extracts initial and structured text features using bi-directional long short-term memory recurrent neural networks and fully connected neural networks, respectively, and fuses these features to generate a target text sequence, improving accuracy and fluency by filtering and correcting initial text content based on structured factual information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automatic text conversion technology is used to generate text articles, then productivity is improved, but manufacturing precision deteriorates due to factual errors and disordered vocabulary
Solution Approach 1:
The patent introduces structured text sequences as an intermediary representation between initial text sequences and target text sequences. This structured representation captures factual information in an organized format, allowing the model to reference and correct itself during generation, thereby improving accuracy without sacrificing productivity
Solution Approach 2:
The patent implements a feedback mechanism where the model generates structured text sequences from initial text, then uses these structured sequences as feedback to guide the generation of final target text sequences. This self-correction process improves manufacturing precision by allowing the model to identify and correct factual errors and vocabulary disorders
2Manufacturing precision
If complex text editing and correction methods are used to improve text accuracy, then manufacturing precision is improved, but device complexity increases
Solution Approach 1:
The patent segments the text generation process into two distinct stages: generating structured text sequences that capture factual information, and then generating target text sequences based on those structured sequences. This segmentation allows each stage to focus on specific aspects of text quality without requiring a single overly complex model
Solution Approach 2:
The patent changes the parameter representation by introducing structured text sequences as an intermediate parameter format. This structured representation transforms unorganized factual information into an organized format that is easier for the model to process and reference, improving editing accuracy without proportionally increasing device complexity
Data Source
AI summary
Embodiments of the present disclosure disclose a text sequence generating method and apparatus, a device and a medium. The method includes: obtaining an initial text sequence, extracting an initial text feature from the initial text sequence; obtaining a structured text sequence, and extracting a structured feature from the structured text sequence, where the structured text sequence is associated with a fact in the initial text sequence; and fusing and generating a target text sequence based on the initial text feature and the structured feature.