Database Reference
In-Depth Information
To resolve this, a BLOB is a package of data with internal structure defined by
the originator and normally requiring custom code to interpret to specify a mapping
from the RDB structure to the RDF graph. There are several tools that facilitate
the publication of a Linked Data view of the RDB by allowing the data publisher
to specify these mappings. These include Virtuoso
12
from Open Link, Triplify
(Auer et al., 2009), R2O (Barrasa and Gómez-Pérez, 2006), and D2R Server.
13
Merea
Maps uses one of these tools to automatically generate the “simplistic” version of
an RDB-to-RDF mapping
14
as it is a good starting point, and then it customizes the
mapping to generate the RDF data corresponding to the RDFS ontology it actually
wants. This includes removing some of the mappings for columns they do not want
to keep (such as the Merea Feature Code) or writing more complex mappings for an
amalgamation of columns (such as Degrees and Minutes of Longitude).
Currently, each RDB-to-RDF tool has its own proprietary way of mapping from
the relational schema to the RDF view. Some allow custom mappings, while others
are limited to simplistic mappings only.
7.5.3.1 Virtuoso
The Virtuoso RDF Views software from Open Link creates “quad map patterns”
using the Virtuoso Meta Schema language to define mappings from a set of RDB
columns to triples in a specific graph (that is, a “quad” of graph, subject, predicate,
object). Virtuoso's SPARQL-to-SQL translator recognizes triple patterns that refer
to graphs that have an RDB representation and translates these into SQL accord-
ingly. The main purpose of this is to map SPARQL queries onto existing RDBs,
but it also enables integration with a triple store. Virtuoso can process a query for
which some triple patterns can be matched against local or remote relational data
and some matched against locally stored RDF triples. Virtuoso tries to make the
right query compilation decisions based on knowledge of the data and its location,
which is especially important when mixing triples and relational data or when deal-
ing with relational data distributed across many external databases. Virtuoso covers
the whole relational model, including multipart keys and so on.
7. 5 . 3 . 2 Tr i p l i f y
Another tool, Triplify,
15
is a software plug-in to Web applications that maps HTTP
GET requests onto SQL queries and transforms the results into RDF triples. These
triples can be published on the Web in various serializations: RDF/XML, JSON
(Java Script Object Notation), or Linked Data (i.e., resources with resolvable URIs
that are linked together on the Web). The software does not support SPARQL queries,
but does enable the publishing of update logs, to facilitate incremental crawling of
Linked Data sources, as well as an extension to specify provenance. Also of note
is that Triplify has been used to publish data from the OpenStreetMap project as
RDF
16
: a very large (greater than 220 million triples) dataset. However, the drawback
of Triplify's approach is that semantic meaning is encoded in the query itself and is
not easily accessible, or indeed decipherable, outside the Triplify system. This prob-
lem is demonstrated in a query
17
used to enable the LinkedGeoData “near” REST
service that returns RDF for nearby point locations:
