Text Analysis System for Narrative Documents Using Temporal Context
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing text analysis engines lose context when analyzing narrative documents, as they are designed based on part-of-speech decomposition and individual word extraction, failing to utilize the temporal information present in documents like accident reports.
Innovation Solution
A system comprising a text analyzing unit that extracts keywords as facets, determines their time sequence, and sorts words into time axes, with a drawing unit creating charts with nodes and edges representing the relationships between facets, incorporating confidence information in sequence rules.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If text analysis is performed using part-of-speech decomposition and individual word extraction, then word-level analysis capability is improved, but temporal context information is lost
Solution Approach 1:
The patent segments the text into time axes based on temporal keywords, creating distinct temporal segments that preserve the chronological structure of the narrative document while enabling detailed word-level analysis within each segment
Solution Approach 2:
The patent adds a temporal dimension to the text analysis by organizing words along time axes, transforming the analysis from a flat word-level approach to a multi-dimensional structure that incorporates temporal context
2Ease of operation
If keywords are extracted as facets without temporal sequencing, then extraction simplicity is improved, but narrative structure understanding deteriorates
Solution Approach 1:
The patent performs preliminary extraction of temporal keywords and establishes time axes before extracting other facets, creating a structured framework that guides subsequent facet extraction while maintaining simplicity in the overall process
3Measurement precision
If chart visualization includes confidence information for sequence rules, then analysis accuracy is improved, but chart complexity increases
Solution Approach 1:
The patent applies local quality by adding confidence information selectively to specific elements of the chart (edges representing sequence rules) rather than uniformly across the entire visualization, providing enhanced accuracy where needed while maintaining overall chart clarity
Data Source
AI summary
A system including: a text analyzing unit extracting keywords as predetermined facets from text of a document to define a time sequence of a part of the text between the keywords and sort words included in the part into time axes, the time axes being divisions of the time sequence; and a drawing unit drawing a chart in a drawing space, the chart including nodes and an edge, the nodes corresponding to the facets, the nodes being arranged in the drawing space in accordance with a relationship of the time sequence between the facets, the edge being linked to the nodes in accordance with a connection between the facets.