Polyglot Persistent Knowledge Graph for Diverse Data Integration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional knowledge graph technologies lack standardization and are unable to effectively store and process diverse data types such as images and time series data, limiting their application in commercial AI and cognitive systems, and there is a need for a system that can seamlessly integrate disparate data sources and allow non-IT users to query them through a unified interface.
Innovation Solution
A polyglot persistent knowledge graph framework that allows data to be stored in the most suitable location based on its type, using semantic triple stores, property graphs, or big data systems, and provides a drag-and-drop interface for users to create queries across multiple data environments, integrating metadata and path-finding services to retrieve and merge data from various sources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional semantic knowledge bases are used to store diverse data types (images, time series, large files), then data storage capability is improved, but storage overhead and processing complexity increase significantly
Solution Approach 1:
The patent segments the knowledge graph into two distinct layers: a semantic layer for storing structured knowledge (entities, relationships, properties) and a binary layer for storing unstructured data (images, time series, large files). This segmentation allows each layer to use optimization techniques appropriate to its data type, reducing overall system complexity while maintaining versatility across multiple data types.
2Productivity
If custom technologies and alternative approaches (such as property graph model) are adopted for specific applications, then application performance is improved, but standardization across enterprises is lost
Solution Approach 1:
The patent creates a universal knowledge graph framework that can handle multiple data types and query patterns through a single standardized interface. The system supports both semantic queries (using RDF/SPARQL) and property graph queries (using custom query languages) while maintaining a unified storage and processing architecture, enabling enterprises to deploy consistent knowledge graph solutions across diverse applications without sacrificing performance optimization.
3Reliability
If all data is stored in semantic triple stores to maintain data integrity and semantics, then data consistency is improved, but storage efficiency and query performance for binary data deteriorate
Solution Approach 1:
The patent introduces an intermediary layer that sits between the binary data storage and the semantic query interface. This intermediary handles the conversion and coordination between binary data representations and semantic knowledge graph structures, allowing binary data to be stored efficiently in optimized formats while still enabling consistent semantic queries through the intermediary translation layer.
4Quantity of substance
If knowledge graphs are expanded to include millions to billions of entities and facts, then information coverage is improved, but system complexity and query processing time increase
Solution Approach 1:
The patent adds a temporal dimension to knowledge graph queries by introducing time-aware operators and functions. This allows queries to efficiently filter and aggregate data based on time ranges, reducing the effective search space from the entire knowledge graph to only relevant time-bounded subsets, thereby maintaining fast query performance even as knowledge graph size grows to billions of entities and facts.
Data Source
AI summary
A method of augmenting a semantic query of multiple external data sources including receiving a request to search a data store for fields-of-interest designated by a user, applying path-finding technique(s) to identify connections between the fields-of-interest in a semantic model, generating a query based on the connections, intercepting the executing query to determine if data for the fields-of-interest are contained in an external data store, if so identifying an external data service to retrieve external data, executing a semantic query on a triple store, fusing results from the semantic query with the retrieved external data, and providing the fused results to the user computing device. A system and a non-transitory computer readable medium are also disclosed.