Overview Archive

Owned by Adam Milward
Jul 31, 2017
2 min read
The MDR is a toolkit for managing metadata:
  • It provides users with contextual information that relates to data elements and models
  • It enables data managers to manage data quality and versioning across a wide variety of information systems.
  • It maps relationships between legacy data-points and new data standards
  • It facilitates metadata re-use and thus enhanced data integration
  • It helps in auditing the datasets, their usage and thus the security of an enterprise

Internally the MDR represents data as a tree model of entities, which are related to each other in a variety of different ways.

The core entities themselves are:

  1. Data Models
    Represent datasets in the metadata registry, it is simply a way of grouping together different models, identifying all the models and data elements together, referencing them and versioning them.
  2. Data Classes
    Represent concepts and are collections of models and data elements, they can be related using a hierarchical relationship or by containment.
  3. Data Elements
    An atomic concept, which will be directly related to one Value Domain which defines how this data element is implemented in this system.
  4. Data Types
    These define how a concept is represented and what values it will take in a given system, in ISO11179 the concept is the given the label of value domain, we prefer to call them simply DataTypes in common with the Computer Science community.
    These artefacts define the type if a DataItem within the system (or technical space) in question, so an Integer will have a particular definition in an XML file, and another definition in a Java program. It may be an enumerated type, say a particular set of strings (Male:Female:Unknown) or it may be a set of integers which represent a particular set of values. The same enumeration may be used in different value domains to represent different data concepts.
  5. Measurement Units
    These are included for reference, since standard units such as SI units crop up in many different systems, and thus it is convenient to reference them in the DataType.

The "Catalogue Elements" can be related in a variety of ways at present the MDR caters for the following relationship types:

  • containment - for instance a Model can contain a number of Data Elements
  • base - for instance one Model can be based on another Model
  • hierarchy - for instance one Model can be a child of another Model
  • relatedTo - this is a catch-all relationship
  • synonym - a data element(say 'firstName' in one Model may be called by the same term in another Model, but may be used and implemented differently)
  • favourite - this is a temporary relationship used to build a collection of Models and Data Elements which can then be used to build a new Model or dataset
  • classification - this is a relationship between all entities which have a particular classification