Metadata
“Structured, encoded data that describe characteristics of information-bearing entities to support in the identification, discovery, assessment, and management of the described entities”
Anything that you can know about the data is metadata.
Three types of metadata
- Business Metadata
- Technical Metadata
- Operational Metadata
Business Metadata (3)
- Focuses on content and condition of data from a business perspective.
- Falls under the conceptual data abstraction level.
- E.g., who can be a client of a business unit
Technical Metadata (3)
- “Technical metadata provides information about the technical details of data, the systems that store data, and the processes that move it within and between systems” (DMBOK).
- Overlap of logical and physical data levels
- E.g., postal code should have 4 numbers and 2 letters.
Operational Metadata (2)
- Describes the details of processing and accessing of data.
- E.g., Data access hierarchies and clearance levels.
Reference Data (2)
- DMBOK’s definition: “Reference data [. . .] is used to relate data in a database to information beyond the boundaries of the organization. […]
- The goal of reference data management is to ensure the organization has access to a complete set of accurate and current values for each concept represented.
Master Data
Provides the best version of the truth about a concept
What do you know about the top down implementation strategie?
Start from birds-eye view and decompose (drill down) each unit to define their metadata.
pro: complying with (new) legislation or starting new strategic initiative (adopting big data analytics)
What do you know about the bottom-up implementation?
Start with individual instances at the lower levels to define data and define abstract classification schemes.
Pro: better for agile or autonomous teams.
Con: may take more time
Masterdata
Provides the best version of the truth about a concept.
Top-down implementation strategy (2)
- Start from birds-eye view and decompose (drill down) each unit to define their metadata.
- Pro when complying with (new) legislation or starting new strategic initiative (adopting big data analytics).
Bottom-up implementation strategy (3)
- Start with individual instances at the lower levels to define data and define abstract classification schemes
- Better for agile or autonomous teams
- Reconciling lower level data definition to integrate with the big picture ‘may take more time’
Ontology (2)
- Branch of philosophy that describes order and structure of reality in broadest sense possible.
- Following representation theory, basing order and structure of IS on order and structure of reality, we would have more faithful representations.
Universal ontology
Universal ontology (like the BWW table with 11 ontology constructs) describes the order and structure of the real world.
Domain ontology
Domain ontology is the set of concepts, relationships between them, and axioms/rules on what can happen and what can exist – the foundation for a knowledge base.
What is OWL and what does it mean?
Ontology web language is a standard language in ontological engineering. Protégé is the most widely used tool for editing ontologies.
Application of Metadata (4)
- Metadata is the foundation of data management as one needs to know about the data assets that are being managed.
- Within organization: what the definition of a certain concept is
- Between organizations: how to reconcile data that might be defined differently
- For security and data quality purposes: we need to know the lineage or provenance of the data
Metadata vs. reference data
- Metadata defines and describes the type of information stored in a database. Such information is a valuable asset in the development and support of information systems
- Reference data, is composed of static facts that are shared by interacting entities
A metadata repository includes (3)
repository = opslagplaats
- Business word-list (i.e., business metadata)
- Data dictionaries (i.e., operational and technical metadata)
- Reference and master data repositories
Metadata Repository Architecture (3 ways)
- Centralized: data definitions from every business unit or entity within the organization is pulled under a universal view
- Distributed: more local autonomy for business units, but less control from a macro level
- Hybrid: business units have their own metadata repositories, and then there is a unified repository at the organization level
Applications of a domain ontology (3)
- Create common understanding of information
- Facilitate interoperability between similar applications
- Use declarative specifications to analyze domain knowledge - Enable repurposing of domain knowledge
- Other organizations can take advantage of formalized knowledge
- Make rules and assumptions of domain more explicit
-Some of the rules are hard-coded in a
programming language, but not easy to
understand for business users
Universal vs Domain ontology
