Knowledge Graph Identity Resolution via Vector Space
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data repositories face challenges in efficiently searching and visualizing data across multiple data stores with different schemas and storage architectures, leading to incomplete search results, duplicate entries, and inaccurate data analytics.
Innovation Solution
A knowledge graph database is used to aggregate and transform data into a common semantic ontology in RDF format, enabling fast search and comparison of entities through a low-dimensional vector space representation, which facilitates the detection and merging of duplicate entries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If data is stored in multiple data stores with different schemas and storage architectures, then data coverage and versatility are improved, but search efficiency and data consistency deteriorate
Solution Approach 1:
The patent introduces a knowledge graph as an intermediary layer between multiple data stores with different schemas and a user query. The knowledge graph acts as a mediator that translates diverse data formats into a unified semantic representation, enabling efficient searching across heterogeneous data sources without requiring direct access to each store's specific schema.
Solution Approach 2:
The knowledge graph provides a universal interface for accessing data from various storage architectures. By representing data from relational databases, flat files, and other formats in a common semantic ontology, the system enables a single search mechanism to query diverse data sources uniformly, improving search efficiency while maintaining data coverage.
2Adaptability or versatility
If data is stored in multiple data stores with different schemas, then data coverage is improved, but duplicate detection accuracy deteriorates
Solution Approach 1:
The patent transforms data from multiple schemas into a standardized semantic representation in the knowledge graph. By changing the representation parameters to a common ontology, the system enables accurate duplicate detection across diverse data sources. Entities that appear different in original stores are normalized to identical representations, making duplicates detectable.
Solution Approach 2:
The knowledge graph serves as an intermediary that normalizes data from different schemas before storage and retrieval. This mediation process ensures that duplicate entities are represented consistently, enabling accurate detection and merging of duplicates while maintaining the ability to access data from multiple original stores.
3Measurement precision
If data is transformed into a common semantic ontology, then search accuracy and duplicate detection are improved, but data transformation complexity increases
Solution Approach 1:
The patent performs data transformation into the common semantic ontology as a preliminary action during data ingestion. By transforming and normalizing data upfront when it enters the knowledge graph, the system avoids complex transformation operations during search and query processing, reducing operational complexity while maintaining high search accuracy.
Solution Approach 2:
The knowledge graph creates a copy of data in a standardized semantic representation rather than working with the original diverse formats. This copying process consolidates transformation complexity into a single normalization step, simplifying subsequent search operations and reducing the complexity of handling diverse data formats during queries.
Data Source
AI summary
A method may include presenting a user interface on a computing device, the user interface including: an input element configured to receive an input of a search concept; a similar concept selection element; and a concept comparison portion; executing a similarity search in a vector-space representation of a knowledge graph using the input; populating the similar concept selection element with a set of a similar concepts based on a result of the executed similarity search; receiving a selection of a comparison concept of the set of similar concepts from the similar concept selection element; generating a predicate comparison visualization in the concept comparison portion based on predicates in the knowledge graph of the comparison concept and the search concept; and presenting the predicate comparison visualization in the concept comparison portion.