Fact Extraction System Using Multi-Stage Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for fact extraction from electronic documents are inefficient, requiring complex linguistic and syntactic analysis of entire documents, which is time-consuming and impractical for large volumes of data.
Innovation Solution
A multi-stage approach that uses a fact-word table to identify factual descriptions, tags parts of speech, and performs detailed analysis only on these descriptions, avoiding extensive processing of the entire document.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex linguistic and syntactic analysis is applied to entire electronic documents to extract facts, then fact extraction accuracy is improved, but processing time increases significantly
Solution Approach 1:
The patent segments the document processing into multiple stages: first identifying candidate factual descriptions using simple pattern matching with fact-words, then applying complex linguistic and syntactic analysis only to these candidates. This segmentation allows the system to maintain high fact extraction accuracy through detailed analysis while reducing overall processing time by limiting complex analysis to a small subset of relevant text portions rather than entire documents.
Solution Approach 2:
The patent performs preliminary identification of candidate factual descriptions using simple fact-word pattern matching before conducting complex linguistic and syntactic analysis. This preliminary action filters out non-factual content early in the process, so that subsequent time-intensive analysis is applied only to promising candidates, thereby maintaining accuracy while significantly reducing total processing time.
2Quantity of substance
If fact extraction is applied to hundreds or thousands of electronic documents, then comprehensive fact coverage is improved, but the time required becomes unacceptable
Solution Approach 1:
The patent divides the fact extraction process into two sequential stages applied to each document: a fast filtering stage using fact-word pattern matching to identify candidate factual descriptions, followed by a more accurate but slower linguistic and syntactic analysis stage applied only to these candidates. This segmentation enables the system to process hundreds or thousands of documents efficiently while maintaining comprehensive fact coverage across the entire document set.
Solution Approach 2:
The patent applies complex linguistic and syntactic analysis to only a partial subset of the document content - specifically, only to candidate factual descriptions identified in the preliminary filtering stage. This partial action approach ensures that comprehensive fact coverage is achieved across all documents while the total processing time remains acceptable, as the expensive analysis is performed on a fraction of the total text.
Data Source
AI summary
Facts are extracted from electronic documents by recognizing factual descriptions using a fact-word table to match to words of the electronic documents. The words of those factual descriptions may be tagged with the appropriate part of speech. More detailed analysis is then performed on those factual descriptions, rather than on the entire electronic document, and particularly to the text in the neighborhood of the fact-word matches. The analysis may involve identifying the linguistic constituents of each phrase and determining the role as either subject or object. Exclusion rules may be applied to eliminate those phrases unlikely to be part of facts, the exclusion rules being based in part on the linguistic constituents. Scoring rules may be applied to remaining phrases, and for those phrases having a score in excess of a threshold, the corresponding sentence part, whole sentence, paragraph, or other document portion may be presented as representing one or more facts.