Compound Noun Phrase Annotation via Segmentation and Mediator
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems fail to annotate electronic text documents with information from compound noun phrases that contain meaningful intervening words, leading to missed annotations of intended phrases.
Innovation Solution
A method that searches for congruent compound noun phrases, determines the preceding word, and annotates the document with a second full term found in a controlled vocabulary, allowing for extraction and annotation of multiple entities from a single compound noun phrase.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If term annotations are disallowed on electronic document texts when there are meaningful intervening words or tokens, then annotation accuracy is maintained, but information from compound noun phrases is lost
Solution Approach 1:
The system segments compound noun phrases into individual terms and annotations by identifying modifiers and heads separately. It divides the compound noun phrase 'epitaxial thin film' into annotatable components, allowing each term to be annotated independently while preserving the relationship between them through structured data fields.
Solution Approach 2:
The patent introduces an intermediary processing layer that analyzes compound noun phrases and extracts meaningful terms. This intermediary system uses natural language processing to identify modifiers and heads, then creates structured annotation data that bridges the gap between the original text and the annotation system, enabling annotation of terms that would otherwise be blocked by intervening words.
2Reliability
If current annotation systems require exact phrase matching in controlled vocabularies, then annotation reliability is improved, but multiple phrases from single compound noun phrases are missed
Solution Approach 1:
The system segments compound noun phrases into individual annotatable terms by identifying modifiers and heads. This allows multiple separate annotations to be created from a single compound noun phrase, with each term independently matched against the controlled vocabulary, thereby improving both reliability through exact matching and productivity by capturing multiple phrases.
Solution Approach 2:
The patent adds a structural dimension to annotation by creating hierarchical relationships between terms. It organizes annotations in a tree structure where compound noun phrases are broken down into constituent terms, allowing the system to match multiple phrases against the controlled vocabulary while maintaining the hierarchical structure that ensures reliability.
3Ease of operation
If compound noun phrases with modifiers are treated as single units, then processing simplicity is maintained, but multiple intended phrases cannot be extracted
Solution Approach 1:
The system automatically segments compound noun phrases into modifiers and heads using natural language processing rules. This segmentation is performed programmatically without requiring manual intervention, maintaining processing simplicity while enabling the extraction of multiple hidden phrases from each compound noun phrase.
Solution Approach 2:
The patent implements self-service annotation where the system automatically analyzes compound noun phrases, identifies modifiers and heads, and creates appropriate annotations without human intervention. The system serves itself by using built-in natural language processing capabilities to extract and annotate multiple phrases from compound noun phrases, eliminating the need for manual processing while capturing hidden information.
Data Source
AI summary
Systems, computer-program products and methods for annotating electronic text documents with multiple entities defined in a controlled vocabulary extracted from a compound noun phrase are disclosed. In one embodiment, a method of annotating an electronic text document includes searching, by a computing device, the electronic text document for instances of congruent compound noun phrases including a head and a modifier. If a congruent compound noun phrase is found, the method further includes determining a preceding word that precedes the modifier of the congruent compound noun phrase, and searching a controlled vocabulary for a second full term having the preceding word and the head of the congruent compound noun phrase. If the second full term is found in the controlled vocabulary, the method further includes annotating the electronic text document with the second full term having the preceding word and the head of the congruent compound noun phrase.