Overview of Data Modeling, Architecture, Master Data Management, Metadata, and Data Quality

Data Modeling and Data Architecture:

Information modeling describes the metadata needed to understand data, processes, and rules related to an enterprise (Figure 1). Information modeling has three main areas:

Data Modeling – Logical data models define business terms and data elements in context, e.g., grouping customer and prospect entities.

Process Modeling – Defines enterprise business processes, using data model entities to describe how data is created or transformed, e.g., the process of a prospect becoming a customer.

Rule Modeling – Describes data governance and compliance policies, specifying quality and management rules that data must follow, e.g., customers must be older than 21 or data older than five years must be archived.

Figure 1

Data modeling is the process where IT and business agree on a common list of business terms (entities), their attribute constraints, and relationships. The ability to maintain and document data models becomes a key capability for organizations to serve various data acquisition needs across critical projects.

There are several forms of data models:

Relational model – used to create online transaction processing (OLTP) systems, typically normalized to third normal form to avoid redundancy.

Dimensional model – used for online analytical processing (OLAP) systems; warehouse design may follow Kimball or Inmon methods, sometimes a hybrid.

Master Data Management

Master Data Management (MDM) includes processes, governance, policies, standards, and tools that consistently define and manage an organization’s critical data, providing a single reference point.

Managed data may include:

Reference data – business objects for transactions and dimensions for analysis.

Analytical data – supports decision making.

Considering that MDM principles aim to keep master data unified and consistent, MDM shares a commonality with Enterprise Information Architecture (EIA): a consistent definition of master data. Ultimately, the process of architecting master data is shared among MDM, Enterprise Information Management (EIM), and EIA. Compared to MDM, the ultimate goal is to create an information management environment that supports the entire information architecture while adding structure and processes to reduce the effort of managing master data.

The following diagram shows the relationship between MDM, EIA, and EIM.

Figure 2

Metadata Management

Metadata provides a reference framework for data. Forrester Research defines metadata as “information that describes or supports the context of data, content, business processes, services, business rules, and policies within an organization’s information system.” For example, Apple’s App Store sells software applications; the applications are data, and metadata includes descriptions, price, user ratings, reviews, and developer.

In a data management environment, several related types of metadata exist:

Technical metadata – technical information about data, such as source table names, column names, and data types (e.g., string, integer).

Business metadata – business context around data, such as names, definitions, owners, and related reference data for business terms.

Operational metadata – information about data usage, such as last update date, access frequency, or last access date.

Metadata management is an end‑to‑end process for creating, enhancing, and maintaining metadata repositories and related processes. It includes establishing processes, mindsets, organization, and capabilities to build a metadata environment. Like BI and MDM, metadata management faces challenges related to business process governance and culture.

The diagram below shows what a metadata repository may contain.

Figure 3

Data Quality Management

Data quality can be viewed as:

The degree to which data reflects the real‑world scenario it describes.

The state of completeness, validity, consistency, timeliness, and accuracy that makes data suitable for a specific purpose.

The sum of characteristics and attributes that determine its ability to meet a given purpose; the overall excellence of data‑related factors.

The processes and techniques that ensure data values meet business requirements and acceptance criteria.

Complete, standard‑based, consistent, accurate, and timestamped.

Data quality management includes establishing and deploying roles, responsibilities, policies, and procedures related to data acquisition, maintenance, dissemination, and disposition. Partnerships between business and technical groups are crucial for success. Business units define data‑governance rules and verify data quality, while IT groups establish and manage the overall environment (architecture, technical infrastructure, systems, and databases) for acquiring, maintaining, disseminating, and disposing of electronic data assets.

The following chart illustrates the data quality management process.

Figure 4