An enterprise ontology (EO) engagement produces a single deliverable, namely a document or software-based encyclopedia (ontology) of the entity types stored in some or all of the enterprise’s servers.
An EO can benefit an IT organization in surprising ways. It helps application and database developers avoid needless data movement and preparation by gathering data directly from primary sources. It helps architects design new systems so they represent entity types in a way that ensures integration with existing sources and replication systems (such as enterprise service buses). And it helps business intelligence and master data management teams gather and distribute shared data efficiently.
Producing an EO may involve three activities:
- setting up an EO server (such as a wiki)
- profiling data sources
- interviewing technical and business staff familiar with the data sources.
Time and headcount requirements scale up with the number of servers and entity types, and with the kinds of entity-type metadata the EO contains. A modest EO (50 entity types in two or three source systems) only requires one consultant for two or three months.
The metadata in the EO deliverable may include the following:
- canonical name
- canonical representation
- logical relations with other entity types
- source systems and their representations
- consumer systems
- replicating systems
- data-quality problems
- data stewards.