Database Reference
In-Depth Information
predicate. Information about the predicate, subject, and object of the triples
may be related to components defined in an existing ontology (which can be
implemented in languages like RDF Schema or OWL). This allows defining
explicitly the semantics of the objects used in the triples and of the assertions
made within these triples. Besides, it allows performing consistency checks
and inferring new information from the information provided in the triples.
RDF information can be stored and queried in an RDF triple store. Over
time, a growing number of metadata catalogs have made use of RDF to store
semantic metadata information. RDF triple stores are often implemented us-
ing a relational database or a hierarchical database as a back end. For ex-
ample, the Jena Semantic Web Toolkit
36
includes functionality to store and
query RDF triples using an Oracle Berkeley DB back end or using a rela-
tional database (PostgreSQL, MySQL, etc.) via a Java database connectiv-
ity (JDBC) interface.
37
Sesame
38
provides a Java framework for storing and
querying RDF triples. The storage and inference layer (SAIL) of Sesame in-
terfaces between RDF functions and the API for various databases, including
relational and object-oriented databases and other RDF stores.
37
Besides basic
querying, these triple stores also implement consistency checking and infer-
ence services that exploit the semantics defined in RDF Schema and OWL
ontologies.
Another technology used in grid environments to deploy metadata catalogs
is the OGSA-DAI (Data Access and Integration) service.
39
The OGSA-DAI
middleware provides a grid service interface that exposes data resources such
as relational or XML databases. Clients of the OGSA-DAI service can store
and query metadata in the back-end database. One example of a metadata
catalog that uses the OGSA-DAI service in its deployment is the Metadata
Catalog Service (MCS),
40
,
41
which provides a set of generic metadata at-
tributes that can be extended with application-specific attributes (described
in Section 12.4.1). MCS is used to store metadata during workflow execu-
tion by the Pegasus workflow management system,
18
,
42
which in turn is used
by a variety of scientific applications, including LIGO,
43
Southern California
Earthquake Context (SCEC)
44
and Montage.
45
Finally, some metadata catalogs are integrated into general-purpose data
management systems, such as the Storage Resource Broker (SRB).
46
SRB
includes an internal metadata catalog called MCAT.
47
SRB supports a logical
namespace that is independent of the physical namespace. Logical files in SRB
can also be aggregated into collections. SRB provides various authentication
mechanisms to access metadata and data within SRB.
12.4.3 Technologies for Provenance Management
The topic of provenance is the focus of many research communities, including
e-Science and grid computing, databases, visualization, digital libraries, Web
technologies, and operating systems. Two surveys by Bose and Frew
48
and
Simmham
49
provide comprehensive overviews of provenance-related concepts,
