Named-entity extraction apparatus with iterative dictionary learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current named-entity extraction methods lack the necessary accuracy, as they rely on manual rules or machine learning, and there is a need for improved recognition techniques to achieve 100% accuracy in extracting and relating entities from document data.
Innovation Solution
A named-entity extraction apparatus that includes a storage device for an extraction dictionary, a document receiving unit, an extraction unit, a designation unit, a relation extraction rule storage device, and a learning unit, which collectively extracts and relates entities by generating learning documents and refining the extraction dictionary based on user input and feedback.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual rules or machine learning methods are used for named-entity extraction, then the extraction process can be automated to some extent, but the accuracy cannot reach 100%
Solution Approach 1:
The patent implements a feedback mechanism where extraction results are fed back into the system for continuous improvement. The extraction unit generates initial extraction results, which are then used to update and refine the extraction dictionary through iterative learning processes, progressively improving accuracy without requiring overly complex manual rule systems.
Solution Approach 2:
The system performs preliminary actions by pre-processing document data and pre-establishing extraction dictionaries before actual extraction. This includes preparing learning data in advance and creating initial extraction rules, which streamlines the subsequent extraction process and improves accuracy without adding complexity during the main extraction operation.
2Measurement precision
If extraction accuracy is improved through more complex methods, then named-entity recognition performance increases, but the processing time and computational resources increase
Solution Approach 1:
The patent applies partial action by focusing extraction efforts on specific named entities and relationships that are most relevant to the query. Rather than processing all possible entities uniformly, the system selectively extracts and processes only the necessary portions of document data, reducing overall processing time while maintaining high accuracy for critical entities.
Solution Approach 2:
The system dynamically adjusts extraction parameters such as confidence thresholds, matching criteria, and processing depth based on the specific document and query characteristics. This allows the system to optimize between accuracy and speed by changing parameters adaptively rather than using fixed complex processing for all cases.
3Reliability
If the extraction dictionary is continuously refined with more rules and patterns, then extraction completeness improves, but the difficulty of maintaining and updating the dictionary increases
Solution Approach 1:
The extraction dictionary maintains itself through automated learning from extraction results and feedback. The system automatically updates and refines the dictionary based on accumulated data and performance metrics, eliminating the need for manual maintenance while continuously improving completeness. The extraction unit and learning system work together to self-optimize the dictionary without external intervention.
4Measurement precision
If user feedback is incorporated to correct extraction results, then extraction accuracy improves, but the interaction complexity between user and system increases
Solution Approach 1:
The system incorporates user feedback through a streamlined mechanism where users can correct extraction results, and these corrections are automatically integrated into the extraction dictionary. The feedback loop is designed to be simple for users while powerfully improving accuracy, as corrections directly update the underlying extraction rules without requiring complex user interfaces or multiple interaction steps.
Data Source
AI summary
In general, according to one embodiment, a named-entity extraction apparatus according to an embodiment includes: an unit extracting, using an extraction dictionary, named entities and relations from extraction document data; a unit designating character strings corresponding to the named entities extracted among character strings in learning document data; a generator generating, by applying a relation extraction rule, a learning document in which relations between named entities belonging to the categories of the relation extraction rule among the named entities designated; and a unit learning the extraction dictionary based on the learning document.