Enterprise Knowledge Graph (EKG)¶
What is an EKG?¶
An Enterprise Knowledge Graph can be seen as "all your data connected", similar to how "the web" connects all web pages, the EKG connects all data objects---and all information that describes the meaning of that data, including in many cases "the knowledge" around that meaning---that exist in all your internal and relevant external systems, potentially at the most granular levels of detail. Since all this data, meaning and knowledge is connected, users can easily find, navigate and query any information with very high levels of accuracy and ease of use. In addition to that, it also serves "use cases"---similar to what end users usually see as "applications"---that leverage the mix of knowledge and data to provide functionality that can go far beyond the current state of the art that traditional technologies allow.
An EKG requires a "platform" to run it, where the term "platform"---or EKG/Platform---is used for the collection of software services that serve the content of the EKG to users and systems. The EKG/Platform can be seen as a "layer" on top of all your other data sources, providing a "holistic view"---from a user perspective---to all data and knowledge, often to its finest detail, that is captured from all your other data sources, connecting all data.
Data Management perspective¶
All data sources that are connected to the EKG are turned into publishers of data, its machine-readable meaning and knowledge in its most normalised form (ideally) creating highly reusable so-called "self-describing datasets" that can support any relevant use case.
Given the strategic importance of creating an EKG that can absorb change and have a very long life-cycle---supporting your strategic use cases---it is critical that your EKG is based on solid industry standards and have no critical components that are based on proprietary technology, avoiding "vendor lock-in" at all costs.
The primary justification for EKG is short and simple: to get your strategic use cases done.
What is a "strategic use case"?¶
One EKG can support an unlimited number of use cases. These use cases are structured as a tree of reusable components, the so-called Use Case Tree (UCT). Each use case represents data or information or knowledge even and behavior. These UCTs are defined and constructed with "the business", your primary stakeholders and budget holders. They capture in non-technical terms which functional requirements your business needs and what the desired Business Outcomes are.
At the top of a Use Case Tree you generally see the Strategic Use Case that your business plans to implement in the long term. Such as true Client 360, Customer Analytics, Enterprise Fraud Detection, real-time Enterprise Risk & Compliance Management, Supply Chain Management, Recommendation Engine / Interest Graph and many others (see catalog).
Depending on how ambitious your use case tree is designed, the EKGF claims that only an EKG can realistically implement those strategic use cases.
The most common and generic reasons for EKG are:
- Achieve a whole new level---or league---of
strategic use cases,
i.e. use cases that cannot be done (properly) today, improving:
- Operational efficiency e.g. automation, cost containment, digitalisation & resiliency
- Control & Oversight e.g. regulatory compliance & governance automation
- Business Innovation e.g. Client 360, improved customer experience & product/market insight
- Organizational capability (no better wat to address "insoluble problems”)
- Become a data-centric operation, leveraging your data, knowledge and staff to become more competitive. Position your organization in the heart of your market and ecosystem.
- Drastically increase data quality: better and faster
decision-making at any level.
- Create much higher levels of trust in the information that your organization provides to executives, employees, customers, suppliers and last but not least: regulators.
- Prepare for AI at a fundamental level.
- Break out of the world of silos. Create a pathway to rationalise your technology landscape and reduce the number of silos.
- Reduce technology cost.
Proposed New Version¶
A knowledge graph is a set of real-world constructs structured in a semantic network of objects, their attributes, and the relationships between objects that can be used to represent the meaning of information. It has a simple semantic structure of Subject – Verb – Object statements, but given sufficient volumes and business coverage of instances of these statements, affords the capability to leverage Semantic AI statistical inference software to traverse and deliver answers to complex business questions. A knowledge graph may also be referred to as Ontology.
An Enterprise Knowledge Graph (EKG) is a knowledge graph with a scope to serve a broad set of information needs across the entire enterprise. The scope is bounded for business information relevant to what the business does, products produced, relevant external knowledge pertaining to the specific industry that the business operates in, and external business environment knowledge that can impact the business. An EKG is built to serve the information needs and usage for stakeholders across the business.
The EKG can be mapped to operational, derived, and external information in other data formats, and to the documents and text that the business uses. It is not necessary to load all operational and external world data into the EKG, it can use data in its current operational supporting state. Knowledge graphs work together with several key technologies that together can achieve significant capabilities in: explainable AI, inferring knowledge from information not directly chained together, improving natural language processing to resolve ambiguity, and quickly finding meaningful information in unstructured data such as text and numerous volumes of documents.