|
Home
People
Publications
Projects
Software
Data
Data model
About attributes
Databases
HEP-Th
Can-o-sleep
Mobile Social Networks
DBLP
News
|
Proximity databases are directed, attributed graphs where nodes
correspond to objects (typically people, places, and things) and
binary links represent the relationships among objects.
Both objects and links in a Proximity database can be structurally
heterogeneous. They can have a variable number of attributes, and a
variable number of values for each attribute.
Attributes
An attribute is a (name, value) pair, and all attributes are set
valued. For example, in a database containing information about
movies, an object representing an actor might have a 'name' attribute
that has multiple values for the actor's given and stage names.
Proximity also permits the creation of multi-column
(multi-dimensional) attributes. For example, a 'location' attribute
might include values for both x and y coordinates. Currently,
multi-column attributes can be represented in Proximity databases, but
their use in Proximity queries or models is not supported.
Representing type information
An important characteristic of Proximity databases is the lack of
a fixed schema that defines object and link types. Type information
can be stored as an attribute value, just like any other attribute.
Users are free to include or not include type information in a
database's object and link attributes and to use whatever name they
chose for an attribute that stores type information.
Special conversion issues
Proximity converts all attribute names to lower case. Attribute
values retain their case. For compatibility with MonetDB, all single
quote, double quote, or newline characters are converted to underscores.
Additional information
|
