Document Summary Generation Using Normalized Appearance Degrees
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing summary generation techniques using multiple algorithms for document summarization often result in unbalanced distribution of element appearance frequencies, leading to biased selection of elements and inadequate accuracy improvements, as they rely on manual weight assignment which is impractical.
Innovation Solution
A summary generating device and method that normalizes appearance degrees extracted by multiple algorithms, filters elements based on these normalized degrees, and selects sentences to generate a summary, using integer linear programming to maximize the sum of normalized appearance degrees while ensuring the selected sentence length is within a predetermined limit.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple algorithms are used to extract elements and select sentences based on appearance frequency, then the diversity of extracted elements increases, but the distribution of appearance frequency becomes unbalanced and specific algorithms are preferentially selected
Solution Approach 1:
The patent applies parameter changes by normalizing the appearance frequency values from different algorithms to a common scale. This normalization process transforms the raw appearance frequencies into comparable normalized values, allowing elements from different algorithms to be fairly evaluated and selected without any single algorithm dominating the selection process.
2Measurement precision
If manual weight assignment is used to balance algorithm contributions, then the accuracy of summary can be improved, but the complexity and impracticality of the method increases
Solution Approach 1:
The patent implements self-service by using the normalization process to automatically balance the contributions of different algorithms. Instead of requiring manual weight assignment, the system autonomously normalizes the appearance frequencies and performs element selection based on these normalized values, making the process both simpler and more practical while maintaining accuracy.
3Ease of operation
If elements are selected based on high appearance frequency from specific algorithms, then the selection process is simplified, but the accuracy and comprehensiveness of the summary decreases
Solution Approach 1:
The patent applies equipotentiality by creating a level playing field through normalization. All algorithms contribute their appearance frequency data on an equal footing, with their values normalized to the same scale. This allows the selection process to remain simple while ensuring that elements from all algorithms have an equal opportunity to be selected, thereby improving both accuracy and comprehensiveness.
Data Source
AI summary
A summary generating device includes at least one processor that is configured to use a plurality of different algorithms, which extract one or more elements from a document and obtain an appearance degree of each of the extracted elements, so as to obtain the elements and the respective appearance degrees of the elements from the document, normalize the obtained appearance degrees for each of the algorithms, select at least one sentence from the document based on the normalized appearance degrees, and generate a summary of the document based on the selected sentence.