Making Data Models Readable - High-Performance Web Databases: Design, Development, and Deployment - page 253

at all. While it is true that this further simplifies the model aesthetically, it
deprives the viewer of very important information about the model's
meaning. Where they are used, it has become a convention to use verbs as
relationship names, but unfortunately verbs, if not completely meaning-
less, are often less than expressive of the full implication and meaning of
the relationship. “Has” is the most commonly used relationship name, and
it tells very little about the true nature of the relationship.
Moreover, when a verb is used, the relationship name often begins to look
like a function name — which is the rightful topic of a different kind of model.
It is preferable to use prepositional phrases, since the preposition, after
all, is the part of speech that specifically addresses relationships. The Ora-
cle Method has a particularly clever way to do this, using a normal but
highly structured sentence to describe an entire relationship. The relation-
ship names are designed to fit into the following structure, reading the rela-
tionship in each direction:
Each
<entity 1>
must be
[or]
may be
<relationship>
one and only one
[or]
one or more
<entity 2>.
The sentence neatly captures not only the relationship itself, but also its
cardinality and optionality. For example, in Exhibit 16, the relationship
shown may be expressed by the two sentences:
1. Each ORDER may be composed of one or more LINE ITEMS .”
2. Each LINE ITEM must be part of one and only one ORDER .”
As described previously, the presence or absence of crow's feet dictates
“one or more” (upper limit many) or “one and only one” (upper limit one).
Exhibit 16-16. Oracle relationship names.