Databases Reference
In-Depth Information
object”, borrowed from the language of object oriented theory, is
meant to emphasize the fact that these things last over time.
Objects should be distinguished from events. Events are the
occasions on which objects are created or destroyed, or on
which objects change their properties or their relationships.
Objects change by means of their involvement in events. A client
changes her name. This event alters a property of the client. A
customer charges a purchase at a store. This event alters the
financial relationship between that customer and that store.
Some objects are relatively static. Documents, like purchase
orders or invoices, are usually created once and then not
modified. Code sets are often represented as tables, and are usu-
ally updated periodically and all at once. Other objects are rela-
tively dynamic. A good example is objects monitored by
instruments, such as atomic nuclei or automobile engines. The
objects that businesses are interested in generally fall in the
mid-range in terms of being static or dynamic.
Both objects and events can be represented in tables. An
event is represented by a transaction, and so a transaction table
is a record of events of the same type, e.g. sales,
inventory
resupply, or claims against policies.
Objects are represented by descriptions of their relevant pro-
perties and relationships, both of which appear in databases as
columns of tables. Each row represents an object, and each col-
umn instance represents either a property of an object, or a rela-
tionship it has to another object or perhaps to an event. In the
case of a many-to-many relationship in which an object is
involved, each instance of the relationship is a row in a separate
table, one that relational theory calls an associative table. These
rows also represent persistent objects, those objects being
relationships which may have their own properties and
relationships that can change over time.
Episodes
Each row in an asserted version table is part of one and only
one episode. An episode is a set of one or more rows representing
the same object. Each row represents that object during a speci-
fied period of time; it is a version of that object, and that period
of time is a period of effective time. If these rows occur one after
the other, i.e. if there are no temporal gaps between them (and
no temporal overlaps either), then these rows belong to the same
episode of that object. But because asserted version tables are
bi-temporal, these rows are also assertions. Episodes are a series
of versions of the same object that are contiguous in effective
