Entity Type Assignment for Fact Repositories
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing fact databases face challenges in assigning entity types to objects with unknown types, due to missing or confounded entity type information, which affects relationship identification and confidence assignment in facts.
Innovation Solution
A method and system that generate entity type models from known objects, create feature sets for objects, and use these models to determine and assign entity types to objects of unknown type, incorporating feature vectors and classification algorithms to improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If entity type information is not available or is confounded in source data, then the completeness and accuracy of entity type assignment deteriorates, but information extraction processes continue without additional verification steps
Solution Approach 1:
The patent introduces an intermediary entity type model that mediates between the object and its entity type assignment. The model is trained on known entity types and used to predict entity types for objects with unknown or confounded type information, effectively acting as a bridge that recovers lost information without requiring direct verification of the original source data.
Solution Approach 2:
The system implements feedback by using the entity type model to assign entity types to objects, then using these assigned entity types to update and refine the model in subsequent iterations. This feedback loop continuously improves entity type assignment accuracy by learning from previously assigned types and applying the improved model to new objects.
2Productivity
If computational methods are used to automatically extract facts from web pages, then productivity increases, but entity type information may be missing or contain errors due to inherent extraction errors
Solution Approach 1:
The entity type model performs self-service by automatically assigning entity types to objects without requiring manual verification or additional information from the source data. The model uses the extracted facts and attributes to independently determine entity types, maintaining the automated extraction process while improving reliability through learned patterns from training data.
3Reliability
If entity type information is used to identify relationships and assign confidence to facts, then the quality of knowledge organization improves, but the system fails when entity type is unknown or confounded
Solution Approach 1:
The system performs preliminary action by training the entity type model in advance on a dataset of objects with known entity types. This pre-trained model can then be applied to new objects to determine entity types before relationship identification and confidence assignment, ensuring that these downstream processes always have entity type information to work with, even when the original source data is incomplete or erroneous.
Data Source
AI summary
A repository contains objects including facts about entities. Objects may be of known or unknown entity type. An entity type assignment engine assigns entity types to objects of unknown entity type. A feature generation module generates a set of features describing the facts included with each object in the repository. An entity type model module generates an entity type model based on the sets of features generated for a subset of objects. An entity type model module generates entity type models, such as a classifier or generative models, based on the sets of features associated with objects of known entity type. An entity type assignment module generates a value based on the sets of features associated with an object of unknown entity type and the entity type model. This value indicates whether the object of unknown entity type is of a known entity type. An object update module stores the object to which the known entity type was assigned in the repository in association with the assigned entity type.