Newsletter Discussion

Abstract

Resource discovery plays a central role in Semantic Web development nowadays. SPARQL - a Protocol And RDF Query Language is aimed at creating a standard in this area. W3C RDF Data Access Working Group (DAWG), who is developing SPARQL, has recently published a new working draft.

SPARQL

Lots of efforts directed towards resource discovery in Semantic Web have been made over the last years. RDF represents the core of this development and allows the handling of resource descriptions. Many RDF generating tools, parsers and storages have already been created. A great deal of tools is being developed or enhanced at present. RDF storage implementations are available on the web as both, commercial and open source software. Some of them are based on relational DBMS; others make use of the possibility to store RDF data coded in RDF/XML in an XML database.

Unfortunately, no query language standard has grown out of these development efforts until now. Implementations based on relational DMBS often introduce their own query languages. Some of these languages syntactically remind us of RDF itself. On the one hand this approach can be considered as a natural way to address information coded in RDF. On the other hand, RDF was designed to be machine-understandable and humans cannot read and write its expressions easily. Software developments based on XML databases provide a possibility to query RDF/XML using XQuery - an XML Query Language and/or XPath - XML Path Language, which are W3C standards for XML retrieval. The problem with this approach is that, compared with XML, RDF provides a more flexible structure. The underlying data model of RDF is a graph whereas the one of XML is a tree. That means that XML representation of an RDF document is not unambiguous. Thus the information obtained by querying particular XML representation of an RDF document can be incomplete.

In order to address these problems, the W3C RDF Data Access Working Group (DAWG)[1] formulated use cases, which an RDF query language must be able to deal with. Resulting from these use cases the working group established critical features that are required for a standard RDF query language. Among the design objectives is the human-friendly syntax of the language [2]. On the April 19^th 2005 W3C published a new version of the SPARQL working draft [3]. SPARQL stands for "Protocol And RDF Query Language" and provides a human readable interface for RDF data retrieval. SPARQL is designed to meet the requirements and design objectives described by DAWG. At first glance, the syntax of SPARQL looks like a mixture of SQL and Prolog. The interface design of SPARQL is quite intuitive.

The development of SPARQL is aimed at creating a standard in the area of Semantic Web resource discovery. Some open source software projects like Jena (a java framework for RDF developed by HP Labs Semantic Web Research) [4] have already claimed their willingness to provide SPARQL implementations within their RDF storages. Developers of Sesam [5], which is an open source RDF database, say in their forum that SPARQL will be supported when it is final [6].

These facts give a feeling that SPARQL in its final version has a good chance to be accepted by the RDF community. The work on SPARQL is not finished yet and there are some issues that are currently under discussion.

References

[1] RDF Data Access Working Group: http://www.w3.org/2001/sw/DataAccess/
[2] RDF Data Access Use Cases and Requirements: http://www.w3.org/TR/rdf-dawg-uc/
[3] SPARQL Query Language for RDF: http://www.w3.org/TR/rdf-sparql-query/
[4] HP Labs Semantic Web Research: http://www.hpl.hp.com/semweb/
[5] OpenRDF.org, Sesame, http://www.openrdf.org/about.jsp
[6] OpenRDF.org, Forum: Sesame usage, Thread: Will Sesame support SPARQL? http://www.openrdf.org/forum/mvnforum/viewthread?thread=502