5.4 Standards and Semantics

Importance of Standards and Semantics

For Metadata Management to succeed across an enterprise, organizations must adopt standardized formats and semantics. Semantic metadata defines the meaning and context of data, ensuring that both humans and machine agents interpret information identically. Without standards, repositories become disjointed "data graveyards" where the same data element is defined, named, and structured in dozens of conflicting ways across different systems. Establishing a unified semantic layer allows organizations to bridge the gap between technical schemas and business concepts, enabling data integration, cataloging, and governance.

---

ISO/IEC 11179 Metadata Registry Standard

The ISO/IEC 11179 standard is the international benchmark for metadata registries. It provides a formal framework for describing, standardizing, and registering data elements to facilitate sharing and reuse. The standard is divided into six parts:

1. Part 1: Framework - Explains the registration concept and the general structure of the standard.
2. Part 2: Classification - Outlines how to classify data elements based on concepts.
3. Part 3: Registry Metamodel and Basic Attributes - Defines the core conceptual model for registering metadata.
4. Part 4: Formulation of Data Definitions - Details rules for creating clear and unambiguous data definitions.
5. Part 5: Naming and Identification Principles - Establishes guidelines for naming data elements consistently.
6. Part 6: Registration - Specifies the procedures for registering and managing data elements.

A central concept in ISO/IEC 11179 is the structure of a Data Element. In this model, data elements are not monolithic; they are built from distinct, reusable parts:

• Object Class: A set of ideas, abstractions, or things in the real world that can be identified with explicit boundaries (e.g., Customer, Employee, Vehicle).
• Property: A characteristic common to all members of an Object Class (e.g., Income, Status, Color).
• Data Element Concept (DEC): The logical combination of an Object Class and a Property, defined independently of its representation (e.g., Customer Income, Employee Status).
• Conceptual Domain: A set of value meanings (categories) that define the semantic scope of a DEC, without specifying the physical representations (e.g., Income Ranges or Employment Status Categories).
• Value Domain: A specific set of permissible values, their physical representations, and datatypes associated with a Data Element (e.g., the set of integers {1, 2, 3} or the string codes {"FT", "PT", "CON"}).
• Representation Class: A classification of the type of value representation used for the Data Element (e.g., Code, Date, Text, Quantity, Identifier).
• Data Element: The actual unit of data that represents the combination of a Data Element Concept and a Value Domain (e.g., Customer Income Code represents Customer Income mapped to a specific set of alphanumeric range codes).

CWM ensures that ETL tools from vendor A, databases from vendor B, and BI tools from vendor C can share a common semantic model of the data pipelines and reporting structures.

Common Warehouse Metamodel (CWM)

The Common Warehouse Metamodel (CWM) is a standard established by the Object Management Group (OMG) for interchanging business intelligence and data warehousing metadata among development tools, repositories, and metadata management applications in distributed environments. CWM is based on three foundational OMG standards:

1. Unified Modeling Language (UML): Used to define the metamodels.
2. XML Metadata Interchange (XMI): Used to serialize and exchange the metadata in XML format.
3. Common Object Request Broker Architecture (CORBA): Used for distributed execution.

CWM models metadata across four main areas:

• Data Resources: Relational databases, non-relational databases, record files, and XML structures.
• Data Analysis: Multidimensional (OLAP) structures, data mining models, information visualization, and business nomenclature (glossaries).
• Data Warehouse Management: ETL transformations, warehouse processes, and execution schedules.
• Foundation: Business information structures, types, and keys.

CWM ensures that ETL tools from vendor A, databases from vendor B, and BI tools from vendor C can share a common semantic model of the data pipelines and reporting structures.

---

Semantic Classification: Taxonomies vs. Ontologies

To navigate and organize complex enterprise metadata, organizations use semantic structures:

Taxonomies

A taxonomy is a hierarchical structure that classifies concepts, entities, or terms into groups based on their characteristics. Relationships in a taxonomy are vertical, representing parent-child or is-a relationships. For example, a taxonomy of financial accounts might classify Checking Account and Savings Account as children of Deposit Account.

• Faceted Taxonomy: Unlike flat hierarchies, a faceted taxonomy allows an object to be classified along multiple independent, parallel dimensions (facets). For instance, a product could be classified by its Material, Color, and Use Case simultaneously.

Ontologies

An ontology is a much richer semantic model that describes concepts, classes, attributes, and the formal relationships between them within a given domain. While taxonomies are limited to hierarchical classifications, ontologies represent complex network-based relationships (e.g., Customer purchased Product, Supplier manufactures Product, Employee reportsTo Manager). Ontologies are essential for building knowledge graphs and semantic search.

Key elements of ontologies include:

• Resource Description Framework (RDF): A standard model for data interchange that represents statements as triples: Subject-Predicate-Object (e.g., John isA Customer).
• Web Ontology Language (OWL): A semantic language designed to represent rich and complex knowledge about groups of things and relations between them.
• Triple Stores: Specialized databases designed to store and query RDF semantic triples using query languages like SPARQL Protocol and RDF Query Language.

---

Business Glossary vs. Data Dictionary

The most common point of confusion in metadata standards is the difference between a Business Glossary and a Data Dictionary. While they are closely related and should be integrated, they serve distinct purposes and audiences:

To achieve maximum value, organizations map Business Glossary terms to Data Dictionary elements. This mapping enables users to search for a business concept (e.g., "Gross Revenue") and automatically discover the exact physical columns across all transactional databases and analytical tables that store that information.

---

Exam Traps and Common Pitfalls

• Trap: Confusing Conceptual Domains and Value Domains - The CDMP exam often tests your ability to distinguish between these two under ISO 11179. A Conceptual Domain defines the meaning (e.g., "valid genders"), while a Value Domain defines the actual allowed values and datatype (e.g., ['M', 'F', 'U'] as character strings).
• Trap: Thinking a Business Glossary is a Technical Directory - A Business Glossary should never contain physical technical references like VARCHAR(50) or table names like TBL_CUST_MGR. It is purely conceptual and business-focused.
• Trap: Recommending Taxonomies for Complex Relationships - If the exam question describes mapping non-hierarchical, complex relationships like "works for", "part of", or "causes", the answer is an Ontology, not a Taxonomy.