Entity Extraction System for Unknown Named Entities
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing information retrieval technologies for named entity extraction in document data, such as natural language processing and named entity recognition, rely on supervised learning techniques, requiring manual intervention and limiting the discovery of unknown named entities, which are not included in training data, and fail to adapt to evolving domains like IT and internet technology.
Innovation Solution
A system utilizing machine learning processes, including a BiLSTM-CRF neural network, for entity extraction and relation extraction to automatically discover and extract unknown named entities from document data, generating a knowledge data graph for natural language processing applications, enabling the discovery of unlabeled and unclassified entities by employing unsupervised and active learning techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If supervised learning techniques are used for named entity extraction, then the system can achieve reliable extraction of known entities from training data, but it cannot discover unknown named entities and requires manual intervention
Solution Approach 1:
The system segments the entity extraction process into two distinct components: a supervised learning component (BiLSTM-CRF) that extracts known entities with high precision, and an unsupervised component that discovers unknown entities through pattern analysis and relationship extraction. This segmentation allows each component to optimize for its specific function without compromising the other.
Solution Approach 2:
The patent introduces an intermediary relationship extraction mechanism that bridges the supervised and unsupervised learning components. This intermediary analyzes relationships between extracted entities and uses relationship patterns to identify and extract unknown entities, serving as a mediator that transfers knowledge from known to unknown entity discovery.
2Measurement precision
If manual intervention is used for entity extraction, then the system can achieve high accuracy for known entities, but productivity decreases and the system cannot adapt to evolving domains
Solution Approach 1:
The system implements self-service through automated unsupervised learning mechanisms that continuously discover new entity types and relationships without human intervention. The relationship extraction component automatically identifies patterns in the data, infers new entity types, and expands the knowledge base autonomously, eliminating the need for continuous manual annotation while maintaining high productivity.
Solution Approach 2:
The patent creates a dynamic extraction system that adapts to evolving domains by continuously learning from new data. The unsupervised learning component dynamically identifies emerging entity types and relationships as they appear in the data stream, allowing the system to automatically adapt to new domains and technologies without retraining or manual configuration.
3Stability of the object's composition
If supervised learning with fixed training data is used, then the system achieves consistent results for training entities, but it fails to adapt to evolving domains like IT and internet technology
Solution Approach 1:
The system implements feedback through its unsupervised learning component that continuously monitors extracted entities and relationships, identifies patterns in emerging data, and feeds this knowledge back into the extraction process. This feedback loop enables the system to maintain consistency for known entities while simultaneously adapting to new domains by learning from actual data patterns rather than relying on static training sets.
Solution Approach 2:
The patent creates a dynamic extraction system that adapts to evolving domains by continuously learning from new data. The unsupervised learning component dynamically identifies emerging entity types and relationships as they appear in the data stream, allowing the system to automatically adapt to new domains and technologies without retraining or manual configuration.
Data Source
AI summary
Systems, computer-implemented methods, and computer program products that can facilitate extraction of named entities from document data to support automation applications are provided. According to an embodiment, a system can comprise a memory that stores computer executable components and a processor that executes the computer executable components stored in the memory. The computer executable components can comprise an entity extraction component that extracts, using a first machine learning process, a first data entity and a second data entity from document data indicative of a textual information. The computer executable components can further comprise a relation extraction component that determines, using a second machine learning process, a relation between the first data entity and the second data entity to generate a knowledge data graph used to execute an application associated with natural language processing for the document data.