Parse Tree-Based Ontology Extraction for Text Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing information extraction systems require reprocessing of entire text documents for new extraction goals and rely on labor-intensive dictionary creation and manual annotation, ignoring syntactic and grammatical information.
Innovation Solution
A system that parses text into parse tree format, identifies verb phrases, and classifies noun phrases as entities and properties based on grammatical relationships, enabling efficient extraction of conceptual relationships without requiring dictionaries or manual annotation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional pipeline-based information extraction systems are used, then extraction can be performed for specific information types, but the entire text must be reprocessed from scratch whenever a new extraction goal emerges or a module is improved
Solution Approach 1:
The system segments the text processing into two independent parts: parsing the text into parse trees (done once) and extracting information based on query conditions (done repeatedly). This allows the extraction step to be optimized for different goals without reprocessing the entire text, resolving the contradiction between adaptability and reprocessing time.
Solution Approach 2:
The text is parsed into parse trees in advance as a preliminary action, creating a structured representation that can be efficiently queried multiple times. This pre-processing step eliminates the need to reparse the entire text for each new extraction goal, reducing reprocessing time while maintaining versatility.
2Productivity
If keyword search is used to locate information, then specific words can be found in documents, but syntactic and grammatical information is ignored
Solution Approach 1:
The system replaces simple keyword-matching mechanisms with parse tree-based syntactic analysis. The parse trees capture grammatical relationships and sentence structure, allowing the system to retrieve information based on both keywords and their syntactic context, thus preventing loss of syntactic information while maintaining search efficiency.
3Measurement precision
If dictionaries are used for named-entity recognition, then known elements can be identified, but creating dictionaries for all syntactic variations is labor-intensive
Solution Approach 1:
The system uses the parse trees themselves to identify entities and properties, eliminating the need for external dictionaries. The grammatical relationships captured in parse trees enable the system to automatically recognize entities and their attributes without manual dictionary creation, reducing labor effort while maintaining recognition accuracy through structural analysis.
4Reliability
If supervised approaches with manually annotated data are used, then models can be trained for named-entity recognition, but the process is labor-intensive
Solution Approach 1:
The system performs unsupervised learning by directly analyzing parse trees to identify entities and properties without requiring manually annotated training data. The parse tree structure provides the necessary information for automatic entity and property extraction, eliminating labor-intensive data annotation while maintaining model reliability through structural grammatical analysis.
Data Source
AI summary
A system for extracting ontological information from a body of text is disclosed. The system parses one or more sentences from the body of text into parse tree format to generate a set of parsed sentences. The system further performs named-entity-recognition by identifying a subset of parsed sentences from the set of parsed sentences. A subset of noun phrases from the subset of parsed sentences are identified and the noun phrases are examined to classify the noun phrases as an entity or as a property. The system also identifies and outputs a conceptual relationship between the entity and the property based at least partially on grammatical relationship of the entity and the property.