3 CREATING THE DATABASE 489
Typical (derived from surveys from values for nominally similar ma-
terials)
Research (preliminary values; work in progress)
Unevaluated (all other data)
3. The Attribute Groups. Having identified the attributes to be included in
the database, they are now grouped by common characteristics in the context of
the end user. Typical groups of attributes are specimen identification, material
composition, specimen condition, specimen preparation, test conditions, me-
chanical properties, electrical properties, physical properties, etc. These groups
of attributes are referred to as relations or tables.
It is important to verify that all the attributes belong to an appropriate relation.
If the database has more than one end use application, the requirements for each
are integrated, while avoiding redundancy. Additional tables may be added to
improve data normalization.
Purely relational databases require the addition of key attributes to each re-
lation that uniquely define each row in the table. These keys are used to link
the tables for the purposes of uniquely identifying a material in the database
and querying information from the database. Hierarchical databases use a com-
bined set of attribute values for a material stored in the database to uniquely
define the material record.
4. The Relation Names for Attribute Groups. A name must be assigned to
each group of attributes using a term that best characterizes the grouping. Re-
lation names are very important in relational databases, as they are specified in
the SQL queries used to retrieve data from the database. Therefore, they must
be unique, descriptive, and unambiguous. Relation names are not as important
in hierarchical databases because they are not typically used in querying, but it
is good practice to maintain their uniqueness.
5. The Relationship between Relations. For databases with a hierarchical
structure, the relationship between the groups of attributes defines the hierarchy
(parent–child relationships) of the database, dictating the structure for the
schema design. This hierarchy provides the path to the property data. The re-
lations should be separated into two groups: those that identify characteristics
for the materials and those that specify properties of the materials. The tables
storing defining characteristics are typically ordered according to the level of
material differentiation the data provides, from the least differentiation (i.e., ma-
terial, test type, component) to the highest (i.e., test conditions, source data).
The relations that contain attributes that define the actual property data are
grouped into tables by type and are typically on the same level at the bottom of
the hierarchy.
For relational databases, the relationship between the groups of attributes
establishes the connection between relations that are logically related to each
other in some form or manner.
15
These relations are used to refine table struc-
tures, minimize redundant data, allow data to be drawn together simultaneously,
and provide for relationship-level integrity.
Hierarchical database relations generally have one-to-many relationships pro-
gressing down the levels of the schema. Relational databases usually have one
or more of three types of relationships: one-to-one, one-to-many, and many-to-
many.