What is Data Modeling?

Not a new concept … the first wave was in the 1960's!
- Databases require schemas – based on data modeling
- Schemas are implementations of data models
- Similar to conceptual database models, but relationships are captured and storage format is not considered although data types are
But let’s be more specific…
- A good data model must
  - Reflect domain knowledge in natural way (not be limited to tables)
  - Be easy to modify (for the scientific and biomedical domains where knowledge changes rapidly)
  - Be application-independent, i.e. avoid implying an implementation format
- Further, since we would like to leverage existing data and share our data, a good data model must
  - Leverage existing ontologies and vocabularies
In the words of the experts at the World Wide Web Consortium (W3C):
- https://www.w3.org/TR/ld-bp/#MODEL

Why is it useful?

Data models are easier for users to understand than schemas
Data models are easier to iterate as subject matter experts and technologists seek a common understanding of system requirements
Data modeling focuses on the data elements and relationships rather than how they are represented
Data modeling allows a decoupling of the model from the schema implementation