Database Reference
In-Depth Information
object being annotated and the
object type
(logical file, collection or
view). The annotation attribute is a string provided by the user. Anno-
tation metadata also includes the distinguished name of the user creating
the annotation and a
timestamp
that records when the annotation was
created.
Creation and transformation history metadata:
These provenance attributes
record process information about how a logical file was created and what
subsequent transformations were performed on the data. This informa-
tion may be used to recreate the data item if it ever gets corrupted, or
the application may decide to recreate the dataset if the cost of recre-
ating it is less than the cost of retrieval.
External catalog metadata:
Because metadata may be spread across multiple
heterogeneous catalogs, these attributes can be used to access external
catalogs.
12.4.2 Technologies for Storing Metadata Management
Metadata catalogs have utilized a variety of underlying technologies, including
relational databases, XML-based databases, grid database services, and RDF
triple stores.
Relational databases are well-suited for metadata repositories in application
domains that have a well-defined metadata ontology that changes relatively
slowly. Relational databases store data in tables and offer good scalability
in both the amount of data they can store and the number of simultaneous
queries they can support. These databases also support the construction of
indexes on particular metadata attributes, which can provide good perfor-
mance for common queries related to those attributes. Scientific collabora-
tions often rely on open source relational databases such as PostgreSQL
29
and MySQL,
30
but some projects use commercial solutions (Oracle, DB2).
Examples of scientific collaborations whose metadata catalogs have used
a relational database include the Laser Interferometer Gravitational-Wave
Observatory (LIGO) project
31
and the Earth System Grid.
32
XML-based databases provide the ability to store and query content stored
in eXtended Markup Language (XML) format. Although some “native” XML
databases store data in XML format, others map XML data to a different
format and use a relational or hierarchical database to store the data. XML
databases can be queried using a variety of languages, such as XPath and
XQuery. Examples of XML databases include the Apache Xindice database
33
and Oracle Berkeley DB XML.
34
The Resource Description Framework (RDF)
35
supports the representation
of graph-based semantic information using a simple data model. An RDF
expression is represented by a set of triples, where each triple contains a
subject, a predicate, and an object. A triple asserts that a relationship exists
between the subject and the object, where the relationship is specified by the
