Relational Knowledge Graphs
People navigate and reason through life with relationships. Knowledge Graphs put facts and relationships in context to enable understanding and analysis.
The world’s biggest companies are using Knowledge Graphs.
They enhance Google search, and power virtual assistants from Google, Apple, and Amazon.
Existing Graph database vendors are positioning themselves around Knowledge Graphs.
Premier consulting firms tell us there is more demand than they can handle for Knowledge Graph projects.
Why is this happening?
Because Knowledge Graphs help teams understand who curates a dataset and who depends upon it being accurate and up-to-date. Knowledge Graphs help bridge existing data silos and promote a coherent understanding across them, with better governance.
Knowledge Graphs enable business to improve their current solutions such as detecting/preventing fraud, or getting a customer-360. They enable discovery of new business opportunities from relationships in connected datasets.
Many organizations currently have hundreds of custom applications, each with their own siloed data. As enterprises shift to a data-centric approach, …
Knowledge Graphs become the foundation of a flexible and extensible system that enables data to be leveraged as an organizational asset.
The RAI Knowledge Graph Management System is designed to provide flexibility, scalability, and performance to enable data-centric enterprise Data Apps to be built around Knowledge Graphs.
Most graph databases are implemented with objects and pointers. A relational Knowledge Graph is built around a relational schema implemented as tables.
The nodes, edges, and attributes of the graph are all first-class citizens in the schema. The relationships that provide meaning also link the tables.
The well-known scalability of the relational model, supporting trillions of rows per table, enables a relational Knowledge Graph to scale to enterprise data.
Relational Knowledge Graphs are built on RAI’s Graph Normal Form. The relationships between elements are implemented as keys that link the tables.
Queries in RAI join tables using their relationship keys. Simply identify a condition that matches one or more nodes, then join them against the related properties.
Contrast this with navigational queries that “hop” from node to node to find relevant attributes.
The relational approach eliminates the single-node vs all-node query distinctions in other graph databases.
Relational queries against GNF use wide n-way JOINs, made efficient by RAI’s “dovetail” JOIN implementation.