The Operational Reference Model:Implementing Dynamic Information Models

To succeed with metadata management, you will need an Operational Reference Model. It is the way to describe the structures and relationships of your business metadata.

The Role and relationship of an Operational Reference Model

Conceptual models, logical models, physical models and canonical models have a specific purpose which is either communication and understanding, or representation of an individual system. Operational Reference Models come into play when organizations require an operational layer which bridges multiple systems and functions and provides a common point of reference or shared language for the enterprise. The Operational Reference Model is similar. to an anti-corruption layer and a semantic layer in its purpose but with an important difference, it's owned and governed by your subject matter experts.

Operational reference model governance

An operational reference model is at its core a set of metadata structured with the help of taxonomies and ontologies. Each part, each domain, is governed by the subject matter experts in the organization in accordance with a federated governance approach.

This approach ensures that IT, Business and Architects all work together within the same framework.

• The role of IT is that of providing a platform (Platform Owner) and/or database for the organization to work on. They also provide a framework and interaction points for the use of the metadata, e.g., integrations.
• The role of Architects (Structure Owner) is that of defining in particular the Ontologies between different taxonomies, but also ensuring that definitions are in place and that attributes are named coherently.
• The role of Domain experts / Domain owners (Structure owner) in a particular area / domain such as Sales, Product and Service is to define the taxonomy levels and which attributes are common within the same domain.
• The role of Subject Matter Experts is that of Data Stewardship where specific meta data is maintained by those who know their specific area. It could be a group of products, a particular service or an industry. Their role is to ensure that all the meta data within their domain is up to date and correct.

Technological requirements

Implementation of an Operational Reference Model requires a database which can manage attribute inheritance, taxonomies (classification structures) and ontologies (relations).

• Attribute inheritance is necessary to ensure that attributes are defined and used in a coherent manner and does not create redundancy and ambiguity in its usage.
• Managing taxonomies is necessary to be able to move between abstract and specific and trace the data lineage between.
• Managing ontologies is required to be able to accurately represent the reality of the organization.

Step by step approach

With the right technology in place the next step would be to:

1. Understand the situation.
2. Design conceptual model.
3. Expand conceptual model to a logical model.
4. Test logical model with real data.
5. Create necessary ontologies.
6. Create and validate ontological relations between data.
7. Verify that the model supports the use case it’s intended to solve.
8. Improve the model.

Example of an Operational Reference Model for Project

Benefits

• An Operational Reference Model will reduce the time spent on new integrations as mapping between systems that use the same language is already done.
• An Operational Reference Model enables metadata quality as it’s owned and managed by the Subject Matter Experts.
• An Operational Reference Model allows for analysis and simulations across abstraction levels with its support for taxonomies.
• An Operational Reference Model allows an organization to create, establish and maintain a vendor independent data layer which reduces the lock in effect of a specific vendor and their data structures.