This blog explores the requirements vital to thorough vetting of knowledge graph platforms. For more detail on these requirements go to the linked white paper The Six Essential Requirements for Scalable Knowledge Graphs.
As a concept, framework, and instrument of production, the knowledge graph phenomenon has become more pervasive throughout the data ecosystem than ever before. Its mounting influence is reflected in enterprise-scale deployments of the most ubiquitous technology companies in existence (including Google, Facebook, and LinkedIn) and the enterprise data strategies of the Fortune 1000, aligning to and influencing the foremost trends impacting the analyst community.
The Gartner Group hails knowledge graphs as a reasonable means of preparing data for machine learning and Artificial Intelligence, as well as an expression of AI itself via graph techniques. It recently announced knowledge graph technology had reached the peak of AI expectations. At the same time, Forrester Research tacitly sanctioned this development by introducing a new recurring vendor report exclusively dedicated to graph data platforms.
Consequently, we see an almost endless array of graph vendors, use cases, and variations of knowledge graphs emerging in the marketplace, each claiming to be authentic representations of this valued approach for integrating data and leveraging advanced analytics. With conflicting definitions abounding, it is essential to disambiguate them to understand what is actually required for enterprise-level deployments. To motivate the discussion, consider our working definition of a knowledge graph:
A Knowledge Graph is a connected graph of data and associated metadata applied to model, integrate and access an organization’s information assets. The knowledge graph represents real-world entities, facts, concepts, and events as well as all the relationships between them yielding a more accurate and more comprehensive representation of an organization’s data.
Thus, scale represents the primary point of distinction. The success of knowledge graphs has advanced beyond the point in which small applications are useful. Today, enterprise-scale is foundational to knowledge graph deployments, which involves supporting up to and beyond:
- Hundreds to thousands of data sources.
- Hundreds of use cases.
- Hundreds of billions of RDF triples (encompassing nodes and edges) across the graph.
- Dozens of users with varying backgrounds building and managing the graph.
Such knowledge graph deployments often fly under the banner of Data Fabric and support advanced analytics use cases. To that end, they drive the following six requirements delivering enterprise-scale for knowledge graphs, including:
- Any Source: True enterprise knowledge graphs are built and maintained from many sources without compromise and with grace, regardless of the sources’ structure variation, format differences, respective data models, or other distinctions at origination.
- Performant Loading and Efficient Storage: Enterprise-scale is impossible to achieve without automating—and expediting—loading source data into the graph and providing options for efficient storing of graph data.
- Flexible Deployments: Knowledge graphs provide limited utility if enterprises are unable to deploy them wherever they’re most advantageous, whether on-premise or in the cloud.
- Real-Time Queries: Entire knowledge graphs must be traversed in real-time to support queries for analytics, data preparation, and data access.
- Easy Interfacing: Knowledge graphs should enable all users and applications (internal and external) to interface with them, regardless of technical capabilities.
- Granular Security: Mainstays of security, data governance, and regulatory compliance are essential to scaling knowledge graphs across organizations.
Not all knowledge graph platforms or solutions fulfill these requirements. Some are less expensive than others or take less time to get started. But for addressing the complexity, breadth, and scalability requirements of leveraging a single investment innumerable times across the enterprise, knowledge graphs must meet these six requirements for companies to build, manage, and query them for any use case.
Dive deeper into these six concepts with the white paper The Six Essential Requirements for Scalable Knowledge Graphs. This white paper can serve as a guide in creating your ideal checklist when considering knowledge graph platforms.