Key Concepts

The Metadata Exchange (MDX) is in essence a metadata registry for managing data, datasets and associated metadata. 

It is built around the idea that metadata is best stored in an aggregated form around the idea of a group of data elements. For instance a traditionally data has been stored in relational databases, these tend to be based around the idea of a group of tables consisting of rows and columns, where the column defines the fields (or data elements) held in the database, and each row is an instance of that data elements.

A database can be captured by the idea of a dataset or dataModel, which is a set of metadata associated with that particular database instance. There may be many data elements or fields, grouped into many tables or data classes in the dataset or data model, but by identifying the data model with a unique (de-referenceable internet friendly identifier) that definition can be stored and managed. All the semantic constraints identified with that set of data can be identified, and the data itself can then be managed.

The dataset has 2 main aspects which we record in the MDX: 

• Descriptive Metadata - and this can be defined by the organisation managing the dataset
• Structural Metadata (i.e. a data dictionary) - which is a record of the structure of the dataset (or database).

Datasets are classified into 3 main types:

• Physical (the most common)
• Logical (which define a structure on which others can be based)
• Conceptual (which have no physical structure but may have a defined ontological structure - as with an OWL ontology)

Datasets in the MDX follow a lifecycle process described here.

Datasets within the MDX are called DataModels.