Learning Outcomes

Conviction

Motivation

• Understanding how metadata provides a key source for understanding the structure of a spatial database, especially for complex multi-feature, normalized, or object-oriented datasets.
• Understanding the value of feature-level metadata as recorded in a GIS database

Skills

General skills:

• Where to find information about database entities and attributes

Advanced skills:

• How to enter database structure information into the metadata content standard

Knowledge

• Understanding the difference between object-oriented databases and relational databases, and differences reflected between FGDC and ISO metadata content standard

Preparatory topics:

Complementary topics:

Vocabulary

Material for this topic

Motivation: basic information about GIS databases

Two basic structures of GIS databases, using DBMS (database management systems):

• Total DBMS solution: all data, both location data (coordinate locations/topology) and attribute data are accessed through the DBMS. With powerful corporate DBMS software (such as Oracle, Sybase, Informix), GIS datasets in this format can be optimized for ultra-fast information retrieval, viewing, and querying. However, true GIS analysis functions are usually absent or limited in this format.
• Mixed solution: some data (usually attribute tables and relationships) are accessed through the DBMS because they fit the model well, but location data are accessed directly by the GIS because they do not fit the DBMS model. Currently, this is the most common DBMS model for GIS software.

Types of database systems

• tabular ("flat file") - data in a single table
• hierarchical
• network
• relational: most popular DBMS model for GIS

The hierarchical, network and relational models all try to deal with the same problem with tabular data: inability to deal with more than one type of object, or with relationships between objects. For instance, a database may need to handle information on aircraft, crew, flights and passengers - four types of records with different attributes, but with relationships between them (e.g. "is booked on" between passenger and flight). This type of organization is referred to as "database normalization". Though it has many advantages, in terms of reducing data redundancy, database normalization also increases the complexity of the database's structure. A data dictionary (which is a subtype of metadata) is critical for interpreting the structure of a normalized database for efficient use.

Types of categorical relationships

• categorical relationships describe the association among individual features in a classification system
• the classification of data is based on the concept of scale of measurement
• there are four scales of measurement:
• nominal --- a qualitative, non-numerical and non-ranking scale that classifies features on intrinsic characteristics
• for example, in a land use classification scheme, polygons can be classified as industrial, commercial, residential, agricultural, public and institutional
• ordinal --- a nominal scale with ranking which differentiates features according to a particular order
• for example, in a land use classification scheme, residential land can be denoted as low density, medium density and high density
• interval --- an ordinal scale with ranking based on numerical values that are recorded with reference to an arbitrary datum
• for example, temperature readings in degrees centigrade are measured with reference to an arbitrary zero (i.e. zero degree temperature does not mean no temperature)
• ratio --- an interval scale with ranking based on numerical values that are measured with reference to an absolute datum
• for example, rainfall data are recorded in mm with reference to an absolute zero (i.e. zero mm rainfall mean no rainfall)
• categorical relationships based on ranking are hierarchical or taxonomic in nature
• this means that data are classified into progressively different levels of detail
• data in the top level are represented by a limited broad basic categories
• data in each basic category are then classified into different sub-categories, which can be further classified into another level if necessary
• the classification of descriptive data is typically based on categorical relationships

Data types associated with attributes: individual attribute records may recorded as integer, real, character, data, or text fields.

Motivation: data dictionaries in metadata

A generic definition of a data dictionary is a catalogue containing information about map features or attributes. The catalog may define data file and element names, sources, accuracy, date of entry or date of update. Within different metadata content formats, the section corresponding to the data dictionary are referenced different ways:

• Entity and Attribute Information section: FGDC content standard (CSDGM)
• Feature Catalogue Information section: ISO content standard

Feature-level metadata

Accuracy is included as one potential component of a data dictionary. Attributes can be defined to keep track of "feature-level metadata": information specific to attributes such as when they were entered, updated, by whom, by what method, quality of method, etc. For instance, road networks often change over time as new roads are built and the data is updated. It is helpful to define attributes for keeping track of which roads were added, when they were added, how they were added, etc. Keeping track of changing features in any dynamic dataset is very helpful in maintaining data quality.
Feature-level metadata is also useful when datasets are comprised of more than one source of information, especially if the sources are of varying data quality. If any concerns about the dataset's quality are brought up in the future, the portions of the dataset whose sources are in question can be easily identified and isolated.

Embedded metadata

Embedded metadata is a form of feature-level metadata. Metadata embedded directly within the data is most frequently designed to describe the accuracy of the elements of the database or the confidance one might place in different elelements of the database.

Example exercises to demonstrate the importance of metadata

Skills: describing entities, attributes, and attribute domains within the Entity and Attribute Information section

There are basically two different methods for describing entities, attributes, and attribute domains within the Entity and Attribute Information section. The first is a textual description, which may be entered under the Overview Description. The Detailed Description is a much more structured method of describing the data dictionary. For simple datasets with only one or two entity types and a limited number of attributes and attribute domains, the Overview section may be sufficient. For more complex databases, it is important to enter information in both the Overview and Detailed Descriptions. The Overview Description should be used as a summary, to explain the structure and relation of different entities; the Detailed Description gives specifics about each of the entity's attributes and attribute domains, including definitions and sources.

In the case of some complex databases, a data dictionary may already have been compiled in a database format. Therefore it is redundant to transfer the information over into the Detailed Description format of the FGDC's content standard. The Overview Description can be used to reference the separate data dictionary, its format, and how to access it.

Some examples of metadata for the Entity and Attribute Information section:

Knowledge: advanced database structures

Learning material for this topic is under construction

Case example: difficulties of representing a dynamic segmentation database in the metadata content standard

Object-oriented databases and the ISO standard