Sentence Extraction System Context Analysis

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing sentence extraction techniques fail to enhance accuracy by not considering context, leading to incorrect extraction of performance factor sentences in financial reports, where relevant sentences may be missed or noise is included.

Innovation Solution

A sentence extraction system that specifies candidate sentences based on extraction conditions, identifies linked sentences, and determines extraction based on the context, using scores and criteria to refine the extraction process, ensuring accurate identification of performance factor sentences.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If sentence extraction is performed based on keyword presence and modification relationship without context consideration, then extraction speed is maintained, but extraction accuracy deteriorates

Engineering Contradiction:
Improvesentence extraction accuracyVSAvoidextraction process complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The extraction process is divided into multiple stages: initial candidate sentence extraction based on keywords, context analysis stage, and final determination stage. This segmentation allows the system to apply simple keyword matching first, then add context analysis only where needed, improving accuracy without proportionally increasing complexity

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system performs preliminary extraction of candidate sentences using simple keyword matching before applying more complex context analysis. This preliminary action filters out obviously irrelevant sentences first, reducing the workload for subsequent context-based verification and improving overall efficiency

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If context is not considered in sentence extraction, then processing speed is maintained, but relevant sentences are omitted and noise is included

Engineering Contradiction:
Improvesentence extraction accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system applies context analysis selectively rather than to all sentences. It performs full context analysis on candidate sentences that meet initial keyword criteria, while skipping sentences that clearly don't match. This partial application of the more time-consuming context analysis method improves accuracy for relevant sentences without proportionally increasing processing time for the entire document

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentUS11238219B2Sentence extraction system, sentence extraction method and information storage medium
Publication Date: 2022.02.01 RAKUTEN GROUP INC
  • US11238219B2 patent drawing
  • US11238219B2 patent drawing
  • US11238219B2 patent drawing

AI summary

At least one processor of a sentence extraction system specifies at least one candidate sentence based on a predetermined extraction condition, the candidate sentence being a candidate for a sentence to be extracted from among a plurality of sentences; specifies at least one linked sentence that has a link to the at least one candidate sentence; and determines whether to extract the at least one linked sentence, based on whether to extract the at least one candidate sentence.