Graph-Based Text Summarization with Keyword Likelihood
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current text summarization techniques, particularly abstractive summarization, face limitations in expressiveness and vocabulary, leading to suboptimal summary generation.
Innovation Solution
A method and system that utilize a graph-based approach with a summary generating device, calculating likelihoods of nodes and paths to select the most relevant keywords and generate summaries, incorporating a beam search algorithm to enhance abstractness and quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If abstractive summarization technique is used to summarize text by abstracting words, then the technique is widely used and easier to implement, but there is a limit of expressiveness and vocabularies in summarization
Solution Approach 1:
The patent combines abstractive summarization (which abstracts words) with generative summarization capabilities (which generates new text through context understanding). The system merges both approaches by using a graph-based model that can both abstract existing keywords and generate new vocabulary, thereby resolving the contradiction between ease of implementation and expressiveness.
Solution Approach 2:
The patent creates a composite summarization approach by integrating multiple techniques: abstractive summarization, generative summarization, and graph-based context modeling. This composite method combines the strengths of different approaches to achieve both implementation feasibility and enhanced expressiveness with expanded vocabulary.
2Adaptability or versatility
If generative summarization technique is used to generate new text through understanding context, then expressiveness is improved, but it is a difficult technique to implement
Solution Approach 1:
The patent segments the complex generative summarization task into manageable components by representing text as a graph structure with nodes and paths. This segmentation allows the system to handle context understanding and text generation in discrete, computable steps, reducing implementation difficulty while maintaining expressiveness.
Solution Approach 2:
The patent introduces a graph-based context model as an intermediary between the input text and the generated summary. This intermediary structure facilitates context understanding by organizing information in a structured graph format, making the complex generative process more manageable and implementable.
3Manufacturing precision
If beam search algorithm is used to select keywords, then the quality and abstractness of summary is improved, but the computational complexity increases
Solution Approach 1:
The patent applies beam search algorithm selectively to key nodes in the graph structure rather than to the entire text processing pipeline. This partial application of beam search maintains summary quality and abstractness while reducing overall computational complexity by limiting the exhaustive search to critical decision points only.
Data Source
AI summary
The invention relates to a method and a system for improving performance of text summarization and has an object of improving performance of a technique for generating a summary from a given paragraph. According to the invention to achieve the object, a method for improving performance of text summarization includes: calculating a first likelihood of each of a plurality of nodes included in a graph corresponding to a natural language-based context; calculating a second likelihood of each of the plurality of nodes by assigning a weight to a first likelihood of a node corresponding to a keyword not presenting in the context among a plurality of keywords corresponding to each of the plurality of nodes; calculating a third likelihood of each of all paths present in the graph based on the second likelihood of each of the plurality of nodes; and generating a summary for the context based on a path having the highest third likelihood among the paths.