In today’s rapidly growing and evolving data environment, it is increasingly difficult for organizations to get maximal value from the full breadth of their data. As more data is produced faster by more sources, it is difficult for analysts, data scientists, and other consumers to be aware of and effectively use data being produced across organizational divisions and domain areas. Impaired discovery and use of data can lead to the direct costs of duplicated effort and wasted time, as well as the opportunity costs of unrealized data-driven insights.
Theconcept of data fabricis increasingly recognized as a key solution to the growing challenges of the modern data ecosystem, utilizing an enterprise-wide data architecture to facilitate discovery and use across the organization. Supporting your data fabric with an ontology–a graph-based semantic model that describes your organization’s data using standardized classes and concepts–amplifies the benefits of the data fabric, allowing your organization to:
At EK we have hadrepeated success drawing on our experience developing ontologies and implementing semantic architectures to provide data fabric solutionsto our clients. These successes have demonstrated tangible ways in which an enterprise ontology empowers a data fabric and magnifies its benefits, and have allowed us to identify concrete steps an organization can take toward designing, piloting, and scaling its own ontology-powered data fabric architecture.
Data fabricis a logical data architecture that connects data assets across an organization and enriches them with standardized semantic metadata. The core of the data fabric is an abstraction layer that uses data virtualization to bring together information about data sources in disparate formats, systems, and locations while applying standardized processes for data governance, privacy, and security. By connecting data as if it were housed together, the data fabric makes it findable and usable without the overhead cost of actually moving the data into a single repository, which could require substantial intermediary processing and the maintenance of redundant infrastructure.
A data fabric can also include components that use knowledge inference, machine learning, or other AI systems to automate integration, orchestration, and metadata management across the systems touched by the data fabric.
Anontologyis a semantic model that defines the concepts in a particular domain area, their attributes, and the relationships between them, encoding them in a graph format that is both machine- and human-readable. Most of the time, an ontology is a generalized model, describing the kinds of entities and relationships that exist in a domain rather than describing specific entities in the domain. For example, an e-commerce business might model the domain area surrounding its fulfillment operations as an ontology with entities such asCustomer,Item, andOrder; attributes such asName,Phone Number,Quantity, andProduct ID; and relationships such asPlacesandContains:
This simple ontology describes the entities involved in a business’s fulfillment operations processes (customers, orders, and items), their attributes (phone numbers, names, quantities, and product IDs), and the relationships between them (a customer places an order, an order contains an item).
In addition to providing a model that bakes in the meaning of data elements, ontologies can also be used to support many advanced AI technologies, includingknowledge graphsandrecommendation engines.
Developing an enterprise ontology to power semantic metadata is central to implementing an effective data fabric. The ontology provides the data fabric with an interoperable schema to standardize, enrich, and reconcile source data, giving users and applications a unified and reliable view of data across disparate data sources.
