Text Summarization via Syntactic Tree Structure Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Graph ranking and deep learning methods for text summarization often ignore key information, fail to consider word order and syntactic structure, leading to non-fluent summaries, and require extensive annotation data for training.
Innovation Solution
A method involving word segmentation, dependency grammar analysis, and tree structure formation to preserve word orders and syntactic structures, allowing for the determination of summary texts without manual annotation or extensive training data, by analyzing each word's removal to form multiple second texts and determining their tree structures for semantic fluency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If graph ranking or deep learning methods are used for text summarization, then automated summary generation is achieved, but key information is ignored and syntactic structure is not preserved
Solution Approach 1:
The patent segments the text into multiple candidate summary texts by removing different words or word sequences from the original text. This segmentation approach allows the system to evaluate multiple possibilities and select the best summary that preserves key information and syntactic structure, rather than relying on a single automated model output.
Solution Approach 2:
The patent employs a feedback mechanism where the quality of each candidate summary text is evaluated based on syntactic structure preservation and key information retention. The system uses this feedback to iteratively improve the summary selection process, comparing candidate summaries against the original text's structural and semantic properties.
2Productivity
If graph ranking or deep learning methods are used for text summarization, then automated processing is enabled, but syntactic structure and word order are not considered
Solution Approach 1:
The patent performs preliminary analysis of the original text's syntactic structure and word relationships before generating summary candidates. By pre-processing the text to identify structural patterns and key grammatical relationships, the system can efficiently evaluate candidate summaries against these predefined structural criteria, maintaining both speed and precision.
Solution Approach 2:
The patent replaces complex deep learning mechanical systems with a more lightweight approach that uses rule-based syntactic analysis and structural comparison. This substitution maintains automated processing while achieving better precision in preserving syntactic structure by explicitly analyzing grammatical relationships rather than relying on black-box model predictions.
3Extent of automation
If graph ranking and deep learning methods are used, then automated summarization is achieved, but extensive annotation data and training are required
Solution Approach 1:
The patent enables the system to evaluate and generate summaries using the text's own structural properties without requiring external annotated training data. The method uses self-contained linguistic analysis and structural comparison techniques that operate directly on the input text, making the system self-sufficient and eliminating the need for extensive annotated corpora for training.
Solution Approach 2:
The patent replaces expensive, resource-intensive deep learning models that require large annotated datasets with a more economical approach using rule-based linguistic analysis and structural evaluation. This substitution reduces the 'cost' in terms of data requirements and computational resources while maintaining automated summarization functionality.
Data Source
AI summary
A text processing method includes: word segmentation is performed on a first text to be processed to determine multiple words of the first text; one word is removed from the first text to obtain one second text; each word is removed from the first text one by one to obtain a second text set; the first text and each second text in the second text set are analyzed respectively to obtain a first tree structure of the first text and a second tree structure of each second text; and a summary text of the first text is determined according to the first tree structure and each second tree structure.