SlideShare a Scribd company logo
ABSTRACT
This paper will present a comparison of current RDF-based
query languages by putting forward query languages that
represent a broad spectrum of architectural approaches to
querying RDF-based data. Query languages reviewed in this
paper include RDQL, SeRQL, SPARQL and XsRQL. A
comparison of these languages will be done with respect to
some of the more desired attributes of an RDF-based query
language. Results of the comparisons will be presented
which will provide insight into which language(s) are the
most mature to date. Finally, we will conclude with some of
the key challenges that lay ahead for any RDF-based query
language to obtain wide spread adoption.
Keywords
RDF, RDFS, Query, Semantic Web
1. INTRODUCTION
The Resource Description Framework (RDF) is a flexible,
extensible architecture to represent metadata information
about World Wide Web resources [12]. As such, RDF
provides a common framework for expressing web resource
metadata so it can be automatically exchanged between
computers without loss of meaning [9]. This framework will
ultimately enable the machine processing of large amounts
of metadata which will help to facilitate the evolution of the
current World Wide Web into the next generation Semantic
Web. With a Semantic Web, web users will no longer need
to be an expert in the domain in which they seek resources in
order to obtain the exact resource needed. Rather, web users
will be able to utilize the metadata present in RDF-based
web resources to enable more efficient information retrieval.
Furthermore, by generalizing the concept of a web resource,
RDF can be used to identify resources identified on the web,
even though the resource may not be directly available on
the web [9].
From a data model standpoint, RDF is based on identifying
web resources (Subjects) utilizing Uniform Resource
Identifiers (URI’s) and describing Subjects with their
respective Properties and Property values. For instance, the
RDF Statement “there is a person identified by web
resourcehttp://www.rh.edu/~kevinh whose full name is
Kevin Hutt” can be represented by the following RDF graph.
.
The graph in Figure 1 can be also represented in an XML
syntax which is the main syntax for exchanging RDF data.
1<?xmlversion=“1.0”?>
2<rdf:RDF xmlns: rdf=”http://www.w3.org/syntax-ns# “
3kxmlns:exterms=http://www.rh.edu/vocab>
4 <rdf:Description rdf :about = http://www.rh.edu/~kevinh>
5 <exterms:fullname>Kevin Hutt</exterms:fullname>
6 </rdf:Description>
7 </rdf:RDF>
Line 1 is an XML declaration which indicates that the
following content is XML. Line 2 indicates that the
following XML content represents RDF. Line 3 declares
another namespace which indicates that URI refs using the
exterms prefix are associated with the vocabulary defined by
the organization at http://www.rh.edu [9]. Line 4 starts with
the Description element which is a container element for the
rdf element being described. Line 4 also contains the about
attribute which describes the subject of the statement. Line 5
describes the property and the property value (Object). Lines
6 and 7 are the closing tags for their respective elements [9]
There is currently no W3C recommendation for an RDF-
based query language; however, a working group was
formed in October 2004 with the goal of producing a
candidate recommendation by April 2005. Prior to the W3C
forming a working group, there have been many RDF-based
query languages proposed and implemented over the last
several years which have including implementations from
industry heavyweights such as Hewlett Packard and Sun
Microsystems. Most notably, Hewlett Packard submitted a
W3C proposal in January 2004 for a query language called
RDQL. A W3C standard for RDF query is an important
milestone in the development of RDF-query languages as it
will likely foster the necessary industry investment and
facilitate widespread adoption. However, as might be
expected with such an important technology, there exists a
wide variety of proposed RDF-query languages each with its
proponents. The open source community offers perhaps a
different approach to standardization for RDF-query. Most
notably the Sesame open source RDF database system has
gained widespread acceptance and will likely be a major
player with respect to standardization efforts.
Standardization promises to be both a politically and
technically challenging endeavor, but very important
nonetheless.
Before presenting any query languages, it is first important
to underscore some of the key motivations for a robust RDF-
based query language:
• A declarative query language will provide easy
access to RDF repositories which will enable
programmers to deploy applications quickly and
efficiently.
A Comparison of RDF Query Languages
Kevin Hutt
The RDF Statement above is a directed graph which is composed of
a Subject node, Property edge and Object node. An RDF
Statement can be thought of as an assertion that a particular
resource (Subject) exists with a specific Property and Object
Value. Both the Subject and Object node are represented by a URI
and an Object node can be represented by a URI or literal value. In
this case, the Object is represented by a text literal.
Figure 1
http://www.
rh.edu/~kev
inh
http://www.rh.ed
u/vocab/fullname
Kevin
Hutt
21st Computer Science Seminar
SE1-T4-1
• A good query language will leverage the semantics
inherent in the RDF schemas and provide more
meaningful results to the user.
• A robust RDF-based query language will be able to
work with many different schema languages since
requiring each resource author to use the same
schema language may not be possible [11].
• An RDF-based query language is also needed in
order to achieve the desired logical and physical
data independence which allows the higher level
language to specify the resources required and
thereby letting the database engine determine how
to store and access the resources [1].
• The query paradigm allows for more intuitive data
access with a shorter learning curve as well as
making access possible in situations where there is
higher operation overhead such as is the case with
protocols such as SOAP [10].
2. QUERY LANGUAGE FEATURES
As aforementioned, there is no consensus about what an
RDF-based query language should offer from a functionality
standpoint. Although there exists a wide body of literature
relating to expressivity within structured data models,
expressivity relating to semi-structured models is still a
relatively new research subject. However, even though many
languages have been implemented with widely varying
syntax, there exist many commonalities in the languages that
have already been implemented with respect to overall
functionality. For the purposes of this paper, a baseline set of
query language features needs to be established to enable
comparison between languages. This paper does not assert
that the language features presented for a basis of
comparison is an authoritative list, rather an informal list
based on common desired features found in languages that
range from the very simple to the more robust.
Before presenting a list of desired features for an RDF –
based query language, it is important to note some of the
peculiarities of the RDF data model which will ultimately
impact the set of operations that should be provided by an
RDF query language [4]. As stated previously, the
underlying data model for RDF is a directed graph which is
composed of a subject node and an object node which are
connected by an edge which identifies the property. The
combination of subject node, object node and property edge
is commonly referred to as a triple which in essence makes
an assertion about a given resource. As a result, the building
blocks of queries are triple patterns which match the user-
supplied triple patterns against the target graph and return a
sub graph wherever the triple pattern finds a match [12]. The
variables in the triple patterns, if any, are bound to the
corresponding subject, predicate or object[12]. The simplest
graph patterns are single triple patterns, but graph patterns
can be combined using various graph operators into more
complicated graph patterns which are essentially ‘anded’
together when queried against the target graph [12].
Whereas RDF allows the user to make simple statements
about resources, RDF users will also want to define
vocabularies used in those statements. Vocabularies allow
users to define the kinds of resources (classes) and which
properties are associated with those classes [9]. RDF Schema
does not necessarily provide a vocabulary of application-
specific classes, rather it provides facilities needed to
describe classes and properties and also indicates which
classes and properties are to be used together [9]. The ability
to leverage the semantics of the RDF schema class
associated with the RDF graph offers query writers powerful
inference tools. And although application-specific
vocabularies are not part of RDF/RDFS, much work has
already been done in this area and vocabularies such as
Dublin Core for web content authoring are beginning to gain
widespread acceptance.
Finally, to further help orient the reader, many of the
languages reviewed in this paper have a SQL-like syntax.
Within SQL, the database is a closed world; the FROM
clause identifies the tables in the database; the WHERE
clause identifies constraints and can be extended with AND
[10]. By analogy, within RDF query, the Web is the database
and the FROM clause identifies the RDF resources available
[10].
2.1 Basic Features
• Generalized Path Expressions: A path
expression syntax for navigating RDF graph
is required which matches a user-input graph
pattern (subject-property-object) against the
underlying graph model. Pattern matching
features typically include:
• Support for searching for
a specific input pattern
within the graph
• Support for substituting
variables in place of a
node or property
• Support for constraining
values using Boolean
expressions
• Value comparison and Data Type
Support: Property values often contain
literals which can be simple strings such as
“joe jones’ or typed literals such as #12-27-
04. An RDF-typed literal consists of a string
literal and its URI reference (datatype URI’s)
which are external to RDF [9].Formally, a
datatype is defined as a set of character
strings known as the lexical space of d, a
non-empty set called the value space of d and
a mapping from the lexical space of d to the
value space of d. This mapping is known as
the lexical-to-value mapping of d [3]. For
instance, let d be a Boolean datatype where
the lexical space = {1,0} (i.e. the list of legal
values to represent the data type) and the
21st Computer Science Seminar
SE1-T4-2
value space = {True, False}. In this case, the
lexical-to-value mapping maps the string 1 to
True and the string 0 to False. In order to
provide adequate support for value
comparison, it is necessary for the query
language to exploit any data types inherent in
the RDF model.
• Closure: In the case of RDF, closure means
that the results of any query operation are
also graphs and not any other data structure.
This means the results of the query can be
used as input to another query and is useful
when one wants to break a larger query into
smaller queries[3].
More Advanced Features:
• Semantic Capabilties: One of the main
advantages of the RDF data model is that it
provides built-in support for semantic
inference via RDFS. Semantic support of a
query language might include:
• The ability to query the
subClass/SubProperties of the result
set
• The ability to query the subClass or
SubProperties of the schema
• The ability for a query language to
support multiple schemas due to
inability to force content providers to
use just one schema [12]
• Optional Values: Most RDF-Query
languages provide path expression
support for exact matching on a
graph. Some languages such as
SPARQL provide support for
specifying optional sections of the
graph similar to an outer join in the
world of SQL. With this
functionality, results can be
returned based on partial matches
of pattern data.
• Aggregate Functions: Aggregate functions
such as MIN, MAX and COUNT have long been
supported in the structured world of SQL and
could provide benefit to the RDF community.
However, very few current languages support
aggregation.
•Advanced Set Operations: Within the relational
database model, several algebraic operations are
implemented which can be combined . These 5
operations form the basis of relational
completeness and include selection, projection,
Cartesian product, set difference and set union [4].
The three basic operations selection, projection and
product are supported by all of the languages
presented in this paper via the path expression;
however, some languages also support set union
and set intersection [4].
3. QUERY LANGUAGE SURVEY
3.1 Overview
A survey of popular RDF query languages recently conducted by
the W3C identified more than 20 languages that are either under
development or have been implemented [6]. Some of the
languages developed provide a basic level of pattern matching
with no inference capabilities while others allow for more
complex semantic processing. Still others were created as
extensions of existing technologies languages such as X-Query to
provide perhaps an easier transition for users of XML-based
query languages. This paper will not perform a complete survey
of all RDF-based query languages, but provide a sampling of
languages that represent a broad cross-section of approaches
including:
• A language with major mainstream
industry support (RDQL) which gives
insight into the level of current
commitment within mainstream
technology companies such as Hewlett
Packard
• A language that has strong W3C support
(SPARQL) which gives insight into the
future direction of standardization efforts
within the W3C
• A language that leverages current XML
approaches (XsRQL) which gives
insight into how RDF-Query can be
added as an extension to other
technologies
• A language with strong collaboration
between the open source community and
industry (SeRQL) and gives insight into
an alternative approach to W3C
standardization
In order to perform a simple side-by-side comparison of
languages listed below, this paper will note whether each query
language provides basic support for the 7 language features
aforementioned. Whereupon a language offers advanced support
for a feature listed below, it will be noted.
1. Value comparison and data type support
2. Generalized path expressions
3. Closure
4. Semantic capabilities
5. Optional values
21st Computer Science Seminar
SE1-T4-3
6. Aggregate functions
7. Advanced set operations (union, intersection)
3.2 RDQL
RDQL was developed by Hewlett Packard and submitted to
the W3C in January 2004. It has been implemented in a
several RDF systems including Jena, RDFStore, Sesame,
PHP XML Classes, 3 Store and RAP-RDF API for PHP [13].
RDQL was derived mainly from SQUISH, an earlier
language. The syntax of RDQL is similar to a SQL-like
select pattern where the select clause allows the projection of
the variables [4]. Based on the limited amount of
functionality offered in RDQL, it would appear that RDQL
was designed to be a simple language. The RDQL approach
“suggests a strategy for possible standardization: an RDQL-
like language could be developed and deployed without
detailed treatment of rule or inference facilities, yet
subsequently we used to query ‘smarter’ RDF services which
make use of inferences licensed by OWL or RDF-rule
semantics” [15].
Syntax Example:
A simple path expression
SELECT ? x
WHERE (?x, http://www.w3.org/1999/02/22-rdf-
syntax,ns#type,http://example.com/someType>)
This select statement matches all statements in the graph that
have a property of http://www.w3.org/1 999/02/22-rdf-
syntax-ns#type and object of http://example.com/someType.
The variable ?x is bound to the subject resource [13].
Table 1: RDQL Scorecard
Data
Type
Supp.
Path
Exp.
Closure Seman
Funct.
Opt
Values
Agg
Funct.
Adv
Set
Ops
Yes X X
No X X X X X
3.3 SPARQL
SPARQL has been designed to meet the requirements and
design goals as described in the W3C RDF Data Access
Working Group (DAWG) document “RDF Data Access Use
Cases and Requirements” [12]. To date, SPARQL has not been
endorsed by the working group and there have been a number
of design issues raised to the working group. Although it is an
incomplete implementation, it represents the latest work of the
W3C and provides insight into the future direction of
standardization efforts within the W3C. It would appear from
the SPARQL Working Draft that the language developers are
following an iterative development process vis-à-vis
incorporating best practices from other languages as well as
direct input from the W3C RDF Data Access Working Group.
As such, it is difficult to say how much additional functionality
SPARQL will have when it is finally complete.
SPARQL offers many of the basic features desired in an RDF-
based query language. SPARQL provides a subset of
operations on plain literals, XSD integers and XSD floats
defined in XQuery and XPath functions and operators such as
comparison of numeric values, functions on string values
casting, comparison of duration, time and date values [12].
SPARQL treats an RDF graph as pure data and is not aware of
any inference that an rdf store may provide [12]. One
interesting feature of SPARQL is the RDF keyword
CONSTRUCT where query results are returned as an RDF
graph specified by an rdf template. Specifically, the
CONSTRUCT keyword returns a single RDF graph formed as
the union of the graph template with variable values obtained
from each query result [12].
Syntax Example:
An example of how SPARQL implements Optional matching
PREFIX foaf: http://xmlns.com/foaf/0.1/
SELECT ?name ?mbox
WHERE (?x foaf:name ? name)
OPTIONAL (?x foaf:mbox ?mbox)
[12]
This query finds the names of people in the dataset and if there is
a mbox property, it will return that as well. The main power of the
OPTIONAL statement is that allows a partial match on the target
graph similar to an outer join in SQL. Without the OPTIONAL
keyword, each triple pattern in a WHERE clause must be satisfied
because the triple patterns are essentially ‘anded’ together to
enable the binding of variables to values.
Table 2: SPARQL Scorecard
Data
Type
Supp.
Path
Exp.
Closure Seman
Funct.
Opt
Values
Agg
Funct.
Adv
Set
Ops
Yes X X X X
No X X X
3.3 SeRQL
SeRQL (pronounced “circle”) is loosely based on several earlier
languages such as RQL, RDQL and N3 [4] and as such represents
a second generation language. SeRQL’s aim is to reconcile ideas
from existing proposals into a proposal that satisfies a lot of these
key requirements [3]. The overall design goal of SeRQL includes
compiling and implementing many of the best features from
earlier languages and delivering a light-weight yet powerful
language [4]. One interesting characteristic of SeRQL is that it
21st Computer Science Seminar
SE1-T4-4
represents a collaboration between industry and the open source
community and seemingly offers a viable alternative to RDF-
query standardization over the W3C process. It is being
developed as a collaboration between the Dutch software
company Aduna and the open source community to address many
of the concerns of the Sesame user community
(www.openrdf.org) [3]. Sesame is an open source RDF database
that offers support for many query languages including SeRQL
which has become the default query language for Sesame.
SeRQL supports many of the basic RDF-query features such as
path expressions, Boolean constraints and advanced features such
as optional matching. Advanced features such as functions for
aggregation are not supported.
Syntax Example:
One of the more interesting features of SeRQL is its support for
advanced path expressions. In situations where one wants to
query =two or more triples with identical subject and predicate,
the subject and predicate do not have to be repeated.
SeRQL: [subj] pred1 [obj1, obj2, obj3]
This expression is equivalent to
[subj1] pred1[obj1]
[subj1] pred1[obj2]
[subj1] pred1[obj3] [3]
SeRQL also supports branched path expressions, which is useful
when multiple properties result from a single node
SeRQL:[subj1] pred1[obj1]
pred2[obj2]
which is equivalent to:
[subj1] pred1[obj1]
[subj1] pred2[obj2] [3]
Another interesting syntax example is how SeRQL supports RDF
reification. Reification is where the subject or object of a
statement is itself a statement. Since it is asyntactic construction,
it can be expressed using basic path expression syntax. [3]
[statemen1t] <rdf:type>[rdf:statement],
[statement1] <rdf:subject>[subj1],
[statement1] <rdf:predicate>[pred1],
[statement1] <rdf:object>[obj1],
[statement1] <rdf:pred2>[obj2],
This is a cumbersome way of dealing with reification. SeRQL
allows one to treat reified statements with a shorthand.
SeRQL:{subj1} pred1 {obj1} pred2 {obj2} [3]
The principles of navigating path expressions are also applied to
navigating class and property hierarchies within SeRQL [3]. To
retrieve the subclasses of a particular class:<my:class1>
{subclass} <rdfs:subClassOf> <my:class1> [3]
Table 3 SeRQL Scorecard
Data
Type
Supp.
Path
Exp.
Closure Sem
Funct.
Opt
Values
Agg
Funct.
Adv
Set
Ops
Yes X X X X X
No X X
3.4 XsRQL
XsRQL is an RDF query language that derives most of syntax and
style from X-Query while leveraging many of the useful features
developed by the W3C XML Query Working Group while
omitting parts of the X-Query specification that are specific to
XML and hence not required in an RDF environment [8]. The
interaction between query evaluation and the data model in
XsRQL is similar to the X-Query: as expressions are evaluated
within XsRQL, they populate instances into the result set from the
target graph. [8]. As the query processor evaluates higher
expressions on the query tree, new items populate the result set or
existing items disappear and the whatever remains at the top of
the query tree are the results of the query [8]. Because of the
relative maturity of XQuery, XsRQL really shines with its support
of built in functions and currently supports string functions such
as chr, empty, exists, string-length as well as aggregate functions
such as count [8].
XsRQL does not follow the familiar SQL syntax of the language
examples previously mentioned in this paper, rather it follows the
slightly abbreviated syntax of XQuery.
Syntax Example
Give me every vcard :Full name in the repository
RDQL SQL-like syntax:
select ?x, ?fname where (?x <http://www.w3.org/2001/vcard-
rdf/3.0#fn> ?fname [8]
In the abbreviated world of XsRQL, the query would look like
this:
@< http://www.w3.org/2001/vcard-rdf/3.0#fn>/* [8]
Another example where we want to find within a datastore triples
where the age of the first partner is less than that of the second
partner
In RDQL:
21st Computer Science Seminar
SE1-T4-5
SELECT ?x, ?y WHERE
(?x:marriedTo ?y)
?(x :age ?xAge)
(?y :age ?yAge)
and ?xAge < ?yAge
In XsRQL:
for $x in * [@<marriedTo>]
for $y in $x/@<marriedTo>/*
return
{$x,@marriedTo>, $y}[8]
This example more fully illustrates how different the
XsRQL language syntax is compared to other RDF query
languages.
Table 4 XsRQL Scorecard
Data
Type
Supp
Path
Exp.
Closure Sem
Funct.
Opt
Values
Agg
Funct.
Adv
Set
Ops
Yes X X X
No X X X X
4. RESULTS
Table 5 : Combined Scorecard
Data
Type
Sup
Path
Exp.
Closur
e
Seman
Funct.
Opt
Values
Agg
Funct.
Ad
v
Set
Ops
SeRQL Yes Yes Yes Yes Yes No No
SPARQ
L
Yes Yes Yes No Yes No No
RDQL Yes Yes No No No No No
XsRQL Yes Yes No No No Yes No
5. CONCLUSIONS
From the basis of this scorecard, SeRQL provides the most
language features to date with SPARQL coming in a close
second. In addition, both SeRQL and SPARQL provide an easy
to read SQL-like syntax. It would appear that SeRQL with strong
support from the open source community and SPARQL with
strong support from the W3C are leading the charge towards
standardization. As previously stated, standardization for RDF
Query is an important step in the evolution of the Semantic Web
as it will likely foster the necessary widespread industry
commitment to make the Semantic Web a success. It will be
interesting to see how quickly the W3C can forward its
standardization efforts with the open source community
progressing as fast as it is. Languages such as XsRQL will likely
struggle to gain a wide foothold due to the arcane syntax of the
language. However, it is possible that one size may not fit all and
one language may not be enough to support the diverse needs of
the RDF community.
Some of the challenges to widespread adoption of RDF Query are
tightly coupled with the challenges of the Semantic Web. As of
January 2005, Peter Norvig, Director of Google’s Search Quality,
reports that there are less than 200,000 documents on the web that
have and .rdf or .owl extension which is approximately .0005%
of the web [16]. What is the point of building RDF query tools if
there are no documents to query? [16]. Another challenge to the
Semantic Web and consequently RDF Query is competing
ontology’s or vocabularies [16]. Content providers within a well-
defined vertical industry might be successful defining ontology’s
based on earlier successes with ANSI X.12 which defines industry
specific EDI transaction sets because customers often force
suppliers to comply [16]. However, it will be difficult to enforce
ontology’s within a broader pool of content providers. Lastly,
RDF Query to provide value, the metadata people affix to content
must be trustworthy and in some cases content providers may be
motivated to provide inaccurate or misleading metadata to lure
search users. In summary, their is a symbiotic relationship
between RDF Query and the larger Semantic Web initiative and it
will be important for RDF Query standardization efforts to align
with the larger issues that the Semantic Web initiative is
encountering in order to gain widespread adoption.
6. REFERENCES
[1] S. Alexaki, V. Christophides, G. Karvounrakis and D.
Plexousakis, “RQL: A Declarative Query Language for RDF”,
Institute of Computer Science, Greece”,[online document]
available at http://www2002.org/CDROM/refereed/329/
[2] J. Angele, D Brickley, S. Decker and J. Saarela, “A Query and
Inference Service for RDF”, January 1998 ”,[online document]
http://www.w3.org/TandS/QL/QL98/pp/queryservice.html
[3] J. Broekstra, A. Kampman, “Se-RQL: An RDF Query and
Transformation Language, Draft”,[online document] available at
http://www.cs.vu.nl/~jbroeks/papers/SeRQL.pdf
[4] J. Broekstra, A. Eberhart, P. Haase, and R. Volx, “A
Comparison of RDF Query Languages”, Proceedings of the Third
International Semantic Web Conference, Hiroshima, Japan,
November 2004”,[online document] available at
http://www.cs.vu.nl/~jbroeks/papers/survey.pdf
[5] K Clark, RDF Data Access Use Cases and Requirements,
October 2004”,[online document] available at
http://www.w3.org/TR/rdf-dawg-uc/
[6] B. Grosof, E. Prud'hommeaux, RDF Query Survey, [online
document], Available at http://www.w3.org/2001/11/13-rdf-
query-rules/
21st Computer Science Seminar
SE1-T4-6
[7] G. Karvounrakis, “RDF Query Languages: A state-of-the-art”,
Institute of Computer Science, Greece online document] available
at http://139.91.183.30:9090/RDF/publications/state.html
[8] H. Katz, “XsRQL:an X-Query-style Query Language for
RDF, A Submission to the RDF Data Access Working Group”
[online document], June 2004, Available at
http://www.fatdog.com/xsrql.html
[9] F. Manola, B.McBride, and E. Miller, Eds., RDF Primer ,
W3C Recomendation ,February 2004 ”,[online document]
available at http://www.w3.org/TR/rdf-primer/
[10] L. Miller 1
, A. Seaborne 2
, A. Reggiori 3
,” Three
Implementations of SquishQL, a Simple RDF Query Language”,
1
ILRT, Bristol University, UK, 2,3
Hewlett-Packard Laboratories,
[online document], available at
http://ilrt.org/discovery/2002/05/squish-
iscw/http://ilrt.org/discovery/2002/05/squish-iscw/
[11] W. Nejdl, W. Siberski and M. Sintek, “Design Issues and
Challenges for RDF and Schema–Based Peer-to-Peer Systems”,
SIGMOD Record, Vol 32, No. 3, September 2003, pp. 41- 46
[12] E. Prud'hommeaux, A. Seaborne, “ SPARQL Query
Language for RDF”, W3C Working Draft 12 October 2, [online
document], Available at http://www.w3.org/tr/rdf-sparql-query/
[13] A. Seaborne, “RDQL, A Query Language for RDF,W3C
Member Submission, [online document], January 2004, Available
at http://www.w3.org/submission/2004/subm-rdql-20040109
[14] W3C, RDF Query and Rules”,[online document] available at
http://www.w3.org/2001/11/13-RDF-Query-Rules/terms
[15] XML-Cover-Pages.Org, Hewlett-Packard Submits Query
Language for RDF (RDQL) to W3C, [online document], January
2004, Available at http://www.xmlcoverpages.org/ni2004-01-14-
a.html
[16] P. Norvig, Semantic,”Web Ontologies: What Works and
What Doesn't”,online document, January 2005, available
http://www.alwayson-
network.com/comments.php?id=7480_0_3_0_C
21st Computer Science Seminar
SE1-T4-7

More Related Content

What's hot

OWL Web Ontology Language Overview
OWL Web Ontology Language OverviewOWL Web Ontology Language Overview
OWL Web Ontology Language Overview
Igor Myroshnichenko
 
Semantic - Based Querying Using Ontology in Relational Database of Library Ma...
Semantic - Based Querying Using Ontology in Relational Database of Library Ma...Semantic - Based Querying Using Ontology in Relational Database of Library Ma...
Semantic - Based Querying Using Ontology in Relational Database of Library Ma...
dannyijwest
 
Semantic web Technology
Semantic web TechnologySemantic web Technology
Semantic web Technology
Abdelrahman Ibrahim
 
Jarrar: OWL -Web Ontology Language
Jarrar: OWL -Web Ontology LanguageJarrar: OWL -Web Ontology Language
Jarrar: OWL -Web Ontology Language
Mustafa Jarrar
 
Jarrar: OWL (Web Ontology Language)
Jarrar: OWL (Web Ontology Language)Jarrar: OWL (Web Ontology Language)
Jarrar: OWL (Web Ontology Language)
Mustafa Jarrar
 
DC-2008 Architecture Forum Open session
DC-2008 Architecture Forum Open sessionDC-2008 Architecture Forum Open session
DC-2008 Architecture Forum Open session
Mikael Nilsson
 
Rdf And Rdf Schema For Ontology Specification
Rdf And Rdf Schema For Ontology SpecificationRdf And Rdf Schema For Ontology Specification
Rdf And Rdf Schema For Ontology Specification
chenjennan
 
ISO MLR semantics
ISO MLR semanticsISO MLR semantics
ISO MLR semantics
Mikael Nilsson
 
Semantic web
Semantic webSemantic web
Semantic web
tariq1352
 
Tools for Next Generation of CMS: XML, RDF, & GRDDL
Tools for Next Generation of CMS: XML, RDF, & GRDDLTools for Next Generation of CMS: XML, RDF, & GRDDL
Tools for Next Generation of CMS: XML, RDF, & GRDDL
Chimezie Ogbuji
 
Modern PHP RDF toolkits: a comparative study
Modern PHP RDF toolkits: a comparative studyModern PHP RDF toolkits: a comparative study
Modern PHP RDF toolkits: a comparative study
Marius Butuc
 
Dublin Core In Practice
Dublin Core In PracticeDublin Core In Practice
Dublin Core In Practice
Marcia Zeng
 
The Web Ontology Language
The Web Ontology LanguageThe Web Ontology Language
The Web Ontology Language
Hector Quintero Casanova
 
Everything you wanted to know about Dublin Core metadata
Everything you wanted to know about Dublin Core metadataEverything you wanted to know about Dublin Core metadata
Everything you wanted to know about Dublin Core metadata
Eduserv Foundation
 
CMDI2RDF
CMDI2RDFCMDI2RDF
CMDI2RDF
CLARIAH
 
Owl web ontology language
Owl  web ontology languageOwl  web ontology language
Owl web ontology language
hassco2011
 
DC Architecture WG Meeting - DC-2006, Mexico
DC Architecture WG Meeting - DC-2006, MexicoDC Architecture WG Meeting - DC-2006, Mexico
DC Architecture WG Meeting - DC-2006, Mexico
Eduserv Foundation
 
20080917 Rev
20080917 Rev20080917 Rev
20080917 Rev
charper
 

What's hot (18)

OWL Web Ontology Language Overview
OWL Web Ontology Language OverviewOWL Web Ontology Language Overview
OWL Web Ontology Language Overview
 
Semantic - Based Querying Using Ontology in Relational Database of Library Ma...
Semantic - Based Querying Using Ontology in Relational Database of Library Ma...Semantic - Based Querying Using Ontology in Relational Database of Library Ma...
Semantic - Based Querying Using Ontology in Relational Database of Library Ma...
 
Semantic web Technology
Semantic web TechnologySemantic web Technology
Semantic web Technology
 
Jarrar: OWL -Web Ontology Language
Jarrar: OWL -Web Ontology LanguageJarrar: OWL -Web Ontology Language
Jarrar: OWL -Web Ontology Language
 
Jarrar: OWL (Web Ontology Language)
Jarrar: OWL (Web Ontology Language)Jarrar: OWL (Web Ontology Language)
Jarrar: OWL (Web Ontology Language)
 
DC-2008 Architecture Forum Open session
DC-2008 Architecture Forum Open sessionDC-2008 Architecture Forum Open session
DC-2008 Architecture Forum Open session
 
Rdf And Rdf Schema For Ontology Specification
Rdf And Rdf Schema For Ontology SpecificationRdf And Rdf Schema For Ontology Specification
Rdf And Rdf Schema For Ontology Specification
 
ISO MLR semantics
ISO MLR semanticsISO MLR semantics
ISO MLR semantics
 
Semantic web
Semantic webSemantic web
Semantic web
 
Tools for Next Generation of CMS: XML, RDF, & GRDDL
Tools for Next Generation of CMS: XML, RDF, & GRDDLTools for Next Generation of CMS: XML, RDF, & GRDDL
Tools for Next Generation of CMS: XML, RDF, & GRDDL
 
Modern PHP RDF toolkits: a comparative study
Modern PHP RDF toolkits: a comparative studyModern PHP RDF toolkits: a comparative study
Modern PHP RDF toolkits: a comparative study
 
Dublin Core In Practice
Dublin Core In PracticeDublin Core In Practice
Dublin Core In Practice
 
The Web Ontology Language
The Web Ontology LanguageThe Web Ontology Language
The Web Ontology Language
 
Everything you wanted to know about Dublin Core metadata
Everything you wanted to know about Dublin Core metadataEverything you wanted to know about Dublin Core metadata
Everything you wanted to know about Dublin Core metadata
 
CMDI2RDF
CMDI2RDFCMDI2RDF
CMDI2RDF
 
Owl web ontology language
Owl  web ontology languageOwl  web ontology language
Owl web ontology language
 
DC Architecture WG Meeting - DC-2006, Mexico
DC Architecture WG Meeting - DC-2006, MexicoDC Architecture WG Meeting - DC-2006, Mexico
DC Architecture WG Meeting - DC-2006, Mexico
 
20080917 Rev
20080917 Rev20080917 Rev
20080917 Rev
 

Viewers also liked

Corre que te como
Corre que te comoCorre que te como
Corre que te como
María José Moya Jiménez
 
Ute karina tinoco vasquez dr.gonzalo remache_fortalecer las capacidades y pot...
Ute karina tinoco vasquez dr.gonzalo remache_fortalecer las capacidades y pot...Ute karina tinoco vasquez dr.gonzalo remache_fortalecer las capacidades y pot...
Ute karina tinoco vasquez dr.gonzalo remache_fortalecer las capacidades y pot...
karmd
 
тренінг на тему
тренінг на темутренінг на тему
тренінг на тему
Людмила Шведенко
 
Publicitatea AquaPark Chisinau
Publicitatea AquaPark ChisinauPublicitatea AquaPark Chisinau
Publicitatea AquaPark Chisinau
Pavel Sidorenco
 
Resume
ResumeResume
Resume
Vijay Nama
 
Comunicación visual tp7
Comunicación visual tp7Comunicación visual tp7
Comunicación visual tp7
Juan Sanvitale
 
Alma
AlmaAlma
Goodrich wallcovering lalika
Goodrich wallcovering lalikaGoodrich wallcovering lalika
Goodrich wallcovering lalika
Imanuel Prabhata
 
Programa completo Ficarq 2015
Programa completo Ficarq 2015Programa completo Ficarq 2015
Programa completo Ficarq 2015
PRODUCCIONES DIMELOAMI
 
публикация алхімова
публикация алхімовапубликация алхімова
публикация алхімоваolya1
 
Groundwater and River Water Interaction on Cikapundung River: Revisited (ICMN...
Groundwater and River Water Interaction on Cikapundung River: Revisited (ICMN...Groundwater and River Water Interaction on Cikapundung River: Revisited (ICMN...
Groundwater and River Water Interaction on Cikapundung River: Revisited (ICMN...
Dasapta Erwin Irawan
 
Personal Branding
Personal BrandingPersonal Branding
Personal Branding
Alessio Fiaschi
 
Turismo con mascotas cualitativo
Turismo con mascotas cualitativoTurismo con mascotas cualitativo
Turismo con mascotas cualitativo
mascotatrip
 
Passive house project
Passive house projectPassive house project
Passive house project
Ky Pichratana
 
CSS3 Flex Layout
CSS3 Flex LayoutCSS3 Flex Layout
CSS3 Flex Layout
Neha Sharma
 
Tutorial 8
Tutorial 8Tutorial 8
Tutorial 8
Drradz Maths
 
01.Key trends in social media
01.Key trends in social media01.Key trends in social media
01.Key trends in social media
Julian Matthews
 
Interoperabilidad Semántica
Interoperabilidad SemánticaInteroperabilidad Semántica
Interoperabilidad Semántica
Alejandro Barros
 
Groundwater and River Water Interaction at Ciromban and Cibeureum Riverbank, ...
Groundwater and River Water Interaction at Ciromban and Cibeureum Riverbank, ...Groundwater and River Water Interaction at Ciromban and Cibeureum Riverbank, ...
Groundwater and River Water Interaction at Ciromban and Cibeureum Riverbank, ...
Dasapta Erwin Irawan
 
新入社員の技術研修に関してありがちな問題を解決するためにやったことやるべきこと
新入社員の技術研修に関してありがちな問題を解決するためにやったことやるべきこと新入社員の技術研修に関してありがちな問題を解決するためにやったことやるべきこと
新入社員の技術研修に関してありがちな問題を解決するためにやったことやるべきこと
pupupopo88
 

Viewers also liked (20)

Corre que te como
Corre que te comoCorre que te como
Corre que te como
 
Ute karina tinoco vasquez dr.gonzalo remache_fortalecer las capacidades y pot...
Ute karina tinoco vasquez dr.gonzalo remache_fortalecer las capacidades y pot...Ute karina tinoco vasquez dr.gonzalo remache_fortalecer las capacidades y pot...
Ute karina tinoco vasquez dr.gonzalo remache_fortalecer las capacidades y pot...
 
тренінг на тему
тренінг на темутренінг на тему
тренінг на тему
 
Publicitatea AquaPark Chisinau
Publicitatea AquaPark ChisinauPublicitatea AquaPark Chisinau
Publicitatea AquaPark Chisinau
 
Resume
ResumeResume
Resume
 
Comunicación visual tp7
Comunicación visual tp7Comunicación visual tp7
Comunicación visual tp7
 
Alma
AlmaAlma
Alma
 
Goodrich wallcovering lalika
Goodrich wallcovering lalikaGoodrich wallcovering lalika
Goodrich wallcovering lalika
 
Programa completo Ficarq 2015
Programa completo Ficarq 2015Programa completo Ficarq 2015
Programa completo Ficarq 2015
 
публикация алхімова
публикация алхімовапубликация алхімова
публикация алхімова
 
Groundwater and River Water Interaction on Cikapundung River: Revisited (ICMN...
Groundwater and River Water Interaction on Cikapundung River: Revisited (ICMN...Groundwater and River Water Interaction on Cikapundung River: Revisited (ICMN...
Groundwater and River Water Interaction on Cikapundung River: Revisited (ICMN...
 
Personal Branding
Personal BrandingPersonal Branding
Personal Branding
 
Turismo con mascotas cualitativo
Turismo con mascotas cualitativoTurismo con mascotas cualitativo
Turismo con mascotas cualitativo
 
Passive house project
Passive house projectPassive house project
Passive house project
 
CSS3 Flex Layout
CSS3 Flex LayoutCSS3 Flex Layout
CSS3 Flex Layout
 
Tutorial 8
Tutorial 8Tutorial 8
Tutorial 8
 
01.Key trends in social media
01.Key trends in social media01.Key trends in social media
01.Key trends in social media
 
Interoperabilidad Semántica
Interoperabilidad SemánticaInteroperabilidad Semántica
Interoperabilidad Semántica
 
Groundwater and River Water Interaction at Ciromban and Cibeureum Riverbank, ...
Groundwater and River Water Interaction at Ciromban and Cibeureum Riverbank, ...Groundwater and River Water Interaction at Ciromban and Cibeureum Riverbank, ...
Groundwater and River Water Interaction at Ciromban and Cibeureum Riverbank, ...
 
新入社員の技術研修に関してありがちな問題を解決するためにやったことやるべきこと
新入社員の技術研修に関してありがちな問題を解決するためにやったことやるべきこと新入社員の技術研修に関してありがちな問題を解決するためにやったことやるべきこと
新入社員の技術研修に関してありがちな問題を解決するためにやったことやるべきこと
 

Similar to semanticweb

Services semantic technology_terminology
Services semantic technology_terminologyServices semantic technology_terminology
Services semantic technology_terminology
Tenforce
 
Linked data HHS 2015
Linked data HHS 2015Linked data HHS 2015
Linked data HHS 2015
Cason Snow
 
.Net and Rdf APIs
.Net and Rdf APIs.Net and Rdf APIs
.Net and Rdf APIs
Recean Denis
 
Semantic web
Semantic web Semantic web
Semantic web
Pallavi Srivastava
 
Matching and merging anonymous terms from web sources
Matching and merging anonymous terms from web sourcesMatching and merging anonymous terms from web sources
Matching and merging anonymous terms from web sources
IJwest
 
RDF and the Semantic Web -- Joanna Pszenicyn
RDF and the Semantic Web -- Joanna PszenicynRDF and the Semantic Web -- Joanna Pszenicyn
RDF and the Semantic Web -- Joanna Pszenicyn
Richard.Sapon-White
 
Semantics
SemanticsSemantics
RDF and Java
RDF and JavaRDF and Java
RDF and Java
Constantin Stan
 
Short Report Bridges performance gap between Relational and RDF
Short Report Bridges performance gap between Relational and RDFShort Report Bridges performance gap between Relational and RDF
Short Report Bridges performance gap between Relational and RDF
Akram Abbasi
 
Deploying PHP applications using Virtuoso as Application Server
Deploying PHP applications using Virtuoso as Application ServerDeploying PHP applications using Virtuoso as Application Server
Deploying PHP applications using Virtuoso as Application Server
webhostingguy
 
Web Spa
Web SpaWeb Spa
XML Bible
XML BibleXML Bible
XML Bible
LiquidHub
 
Extracting Authoring Information Based on Keywords andSemant.docx
Extracting Authoring Information Based on Keywords andSemant.docxExtracting Authoring Information Based on Keywords andSemant.docx
Extracting Authoring Information Based on Keywords andSemant.docx
mydrynan
 
Resource description framework
Resource description frameworkResource description framework
Resource description framework
Stanley Wang
 
Rdf Processing Tools In Java
Rdf Processing Tools In JavaRdf Processing Tools In Java
Rdf Processing Tools In Java
DicusarCorneliu
 
A REVIEW ON RDB TO RDF MAPPING FOR SEMANTIC WEB
A REVIEW ON RDB TO RDF MAPPING FOR SEMANTIC WEB A REVIEW ON RDB TO RDF MAPPING FOR SEMANTIC WEB
A REVIEW ON RDB TO RDF MAPPING FOR SEMANTIC WEB
ijfcstjournal
 
A REVIEW ON RDB TO RDF MAPPING FOR SEMANTIC WEB
A REVIEW ON RDB TO RDF MAPPING FOR SEMANTIC WEB A REVIEW ON RDB TO RDF MAPPING FOR SEMANTIC WEB
A REVIEW ON RDB TO RDF MAPPING FOR SEMANTIC WEB
ijfcstjournal
 
A Review on RDB to RDF Mapping for Semantic Web
A Review on RDB to RDF Mapping for Semantic WebA Review on RDB to RDF Mapping for Semantic Web
A Review on RDB to RDF Mapping for Semantic Web
AEIJjournal2
 
Jpl presentation
Jpl presentationJpl presentation
Jpl presentation
Rama Bastola
 
Jpl presentation
Jpl presentationJpl presentation
Jpl presentation
Rama Bastola
 

Similar to semanticweb (20)

Services semantic technology_terminology
Services semantic technology_terminologyServices semantic technology_terminology
Services semantic technology_terminology
 
Linked data HHS 2015
Linked data HHS 2015Linked data HHS 2015
Linked data HHS 2015
 
.Net and Rdf APIs
.Net and Rdf APIs.Net and Rdf APIs
.Net and Rdf APIs
 
Semantic web
Semantic web Semantic web
Semantic web
 
Matching and merging anonymous terms from web sources
Matching and merging anonymous terms from web sourcesMatching and merging anonymous terms from web sources
Matching and merging anonymous terms from web sources
 
RDF and the Semantic Web -- Joanna Pszenicyn
RDF and the Semantic Web -- Joanna PszenicynRDF and the Semantic Web -- Joanna Pszenicyn
RDF and the Semantic Web -- Joanna Pszenicyn
 
Semantics
SemanticsSemantics
Semantics
 
RDF and Java
RDF and JavaRDF and Java
RDF and Java
 
Short Report Bridges performance gap between Relational and RDF
Short Report Bridges performance gap between Relational and RDFShort Report Bridges performance gap between Relational and RDF
Short Report Bridges performance gap between Relational and RDF
 
Deploying PHP applications using Virtuoso as Application Server
Deploying PHP applications using Virtuoso as Application ServerDeploying PHP applications using Virtuoso as Application Server
Deploying PHP applications using Virtuoso as Application Server
 
Web Spa
Web SpaWeb Spa
Web Spa
 
XML Bible
XML BibleXML Bible
XML Bible
 
Extracting Authoring Information Based on Keywords andSemant.docx
Extracting Authoring Information Based on Keywords andSemant.docxExtracting Authoring Information Based on Keywords andSemant.docx
Extracting Authoring Information Based on Keywords andSemant.docx
 
Resource description framework
Resource description frameworkResource description framework
Resource description framework
 
Rdf Processing Tools In Java
Rdf Processing Tools In JavaRdf Processing Tools In Java
Rdf Processing Tools In Java
 
A REVIEW ON RDB TO RDF MAPPING FOR SEMANTIC WEB
A REVIEW ON RDB TO RDF MAPPING FOR SEMANTIC WEB A REVIEW ON RDB TO RDF MAPPING FOR SEMANTIC WEB
A REVIEW ON RDB TO RDF MAPPING FOR SEMANTIC WEB
 
A REVIEW ON RDB TO RDF MAPPING FOR SEMANTIC WEB
A REVIEW ON RDB TO RDF MAPPING FOR SEMANTIC WEB A REVIEW ON RDB TO RDF MAPPING FOR SEMANTIC WEB
A REVIEW ON RDB TO RDF MAPPING FOR SEMANTIC WEB
 
A Review on RDB to RDF Mapping for Semantic Web
A Review on RDB to RDF Mapping for Semantic WebA Review on RDB to RDF Mapping for Semantic Web
A Review on RDB to RDF Mapping for Semantic Web
 
Jpl presentation
Jpl presentationJpl presentation
Jpl presentation
 
Jpl presentation
Jpl presentationJpl presentation
Jpl presentation
 

semanticweb

  • 1. ABSTRACT This paper will present a comparison of current RDF-based query languages by putting forward query languages that represent a broad spectrum of architectural approaches to querying RDF-based data. Query languages reviewed in this paper include RDQL, SeRQL, SPARQL and XsRQL. A comparison of these languages will be done with respect to some of the more desired attributes of an RDF-based query language. Results of the comparisons will be presented which will provide insight into which language(s) are the most mature to date. Finally, we will conclude with some of the key challenges that lay ahead for any RDF-based query language to obtain wide spread adoption. Keywords RDF, RDFS, Query, Semantic Web 1. INTRODUCTION The Resource Description Framework (RDF) is a flexible, extensible architecture to represent metadata information about World Wide Web resources [12]. As such, RDF provides a common framework for expressing web resource metadata so it can be automatically exchanged between computers without loss of meaning [9]. This framework will ultimately enable the machine processing of large amounts of metadata which will help to facilitate the evolution of the current World Wide Web into the next generation Semantic Web. With a Semantic Web, web users will no longer need to be an expert in the domain in which they seek resources in order to obtain the exact resource needed. Rather, web users will be able to utilize the metadata present in RDF-based web resources to enable more efficient information retrieval. Furthermore, by generalizing the concept of a web resource, RDF can be used to identify resources identified on the web, even though the resource may not be directly available on the web [9]. From a data model standpoint, RDF is based on identifying web resources (Subjects) utilizing Uniform Resource Identifiers (URI’s) and describing Subjects with their respective Properties and Property values. For instance, the RDF Statement “there is a person identified by web resourcehttp://www.rh.edu/~kevinh whose full name is Kevin Hutt” can be represented by the following RDF graph. . The graph in Figure 1 can be also represented in an XML syntax which is the main syntax for exchanging RDF data. 1<?xmlversion=“1.0”?> 2<rdf:RDF xmlns: rdf=”http://www.w3.org/syntax-ns# “ 3kxmlns:exterms=http://www.rh.edu/vocab> 4 <rdf:Description rdf :about = http://www.rh.edu/~kevinh> 5 <exterms:fullname>Kevin Hutt</exterms:fullname> 6 </rdf:Description> 7 </rdf:RDF> Line 1 is an XML declaration which indicates that the following content is XML. Line 2 indicates that the following XML content represents RDF. Line 3 declares another namespace which indicates that URI refs using the exterms prefix are associated with the vocabulary defined by the organization at http://www.rh.edu [9]. Line 4 starts with the Description element which is a container element for the rdf element being described. Line 4 also contains the about attribute which describes the subject of the statement. Line 5 describes the property and the property value (Object). Lines 6 and 7 are the closing tags for their respective elements [9] There is currently no W3C recommendation for an RDF- based query language; however, a working group was formed in October 2004 with the goal of producing a candidate recommendation by April 2005. Prior to the W3C forming a working group, there have been many RDF-based query languages proposed and implemented over the last several years which have including implementations from industry heavyweights such as Hewlett Packard and Sun Microsystems. Most notably, Hewlett Packard submitted a W3C proposal in January 2004 for a query language called RDQL. A W3C standard for RDF query is an important milestone in the development of RDF-query languages as it will likely foster the necessary industry investment and facilitate widespread adoption. However, as might be expected with such an important technology, there exists a wide variety of proposed RDF-query languages each with its proponents. The open source community offers perhaps a different approach to standardization for RDF-query. Most notably the Sesame open source RDF database system has gained widespread acceptance and will likely be a major player with respect to standardization efforts. Standardization promises to be both a politically and technically challenging endeavor, but very important nonetheless. Before presenting any query languages, it is first important to underscore some of the key motivations for a robust RDF- based query language: • A declarative query language will provide easy access to RDF repositories which will enable programmers to deploy applications quickly and efficiently. A Comparison of RDF Query Languages Kevin Hutt The RDF Statement above is a directed graph which is composed of a Subject node, Property edge and Object node. An RDF Statement can be thought of as an assertion that a particular resource (Subject) exists with a specific Property and Object Value. Both the Subject and Object node are represented by a URI and an Object node can be represented by a URI or literal value. In this case, the Object is represented by a text literal. Figure 1 http://www. rh.edu/~kev inh http://www.rh.ed u/vocab/fullname Kevin Hutt 21st Computer Science Seminar SE1-T4-1
  • 2. • A good query language will leverage the semantics inherent in the RDF schemas and provide more meaningful results to the user. • A robust RDF-based query language will be able to work with many different schema languages since requiring each resource author to use the same schema language may not be possible [11]. • An RDF-based query language is also needed in order to achieve the desired logical and physical data independence which allows the higher level language to specify the resources required and thereby letting the database engine determine how to store and access the resources [1]. • The query paradigm allows for more intuitive data access with a shorter learning curve as well as making access possible in situations where there is higher operation overhead such as is the case with protocols such as SOAP [10]. 2. QUERY LANGUAGE FEATURES As aforementioned, there is no consensus about what an RDF-based query language should offer from a functionality standpoint. Although there exists a wide body of literature relating to expressivity within structured data models, expressivity relating to semi-structured models is still a relatively new research subject. However, even though many languages have been implemented with widely varying syntax, there exist many commonalities in the languages that have already been implemented with respect to overall functionality. For the purposes of this paper, a baseline set of query language features needs to be established to enable comparison between languages. This paper does not assert that the language features presented for a basis of comparison is an authoritative list, rather an informal list based on common desired features found in languages that range from the very simple to the more robust. Before presenting a list of desired features for an RDF – based query language, it is important to note some of the peculiarities of the RDF data model which will ultimately impact the set of operations that should be provided by an RDF query language [4]. As stated previously, the underlying data model for RDF is a directed graph which is composed of a subject node and an object node which are connected by an edge which identifies the property. The combination of subject node, object node and property edge is commonly referred to as a triple which in essence makes an assertion about a given resource. As a result, the building blocks of queries are triple patterns which match the user- supplied triple patterns against the target graph and return a sub graph wherever the triple pattern finds a match [12]. The variables in the triple patterns, if any, are bound to the corresponding subject, predicate or object[12]. The simplest graph patterns are single triple patterns, but graph patterns can be combined using various graph operators into more complicated graph patterns which are essentially ‘anded’ together when queried against the target graph [12]. Whereas RDF allows the user to make simple statements about resources, RDF users will also want to define vocabularies used in those statements. Vocabularies allow users to define the kinds of resources (classes) and which properties are associated with those classes [9]. RDF Schema does not necessarily provide a vocabulary of application- specific classes, rather it provides facilities needed to describe classes and properties and also indicates which classes and properties are to be used together [9]. The ability to leverage the semantics of the RDF schema class associated with the RDF graph offers query writers powerful inference tools. And although application-specific vocabularies are not part of RDF/RDFS, much work has already been done in this area and vocabularies such as Dublin Core for web content authoring are beginning to gain widespread acceptance. Finally, to further help orient the reader, many of the languages reviewed in this paper have a SQL-like syntax. Within SQL, the database is a closed world; the FROM clause identifies the tables in the database; the WHERE clause identifies constraints and can be extended with AND [10]. By analogy, within RDF query, the Web is the database and the FROM clause identifies the RDF resources available [10]. 2.1 Basic Features • Generalized Path Expressions: A path expression syntax for navigating RDF graph is required which matches a user-input graph pattern (subject-property-object) against the underlying graph model. Pattern matching features typically include: • Support for searching for a specific input pattern within the graph • Support for substituting variables in place of a node or property • Support for constraining values using Boolean expressions • Value comparison and Data Type Support: Property values often contain literals which can be simple strings such as “joe jones’ or typed literals such as #12-27- 04. An RDF-typed literal consists of a string literal and its URI reference (datatype URI’s) which are external to RDF [9].Formally, a datatype is defined as a set of character strings known as the lexical space of d, a non-empty set called the value space of d and a mapping from the lexical space of d to the value space of d. This mapping is known as the lexical-to-value mapping of d [3]. For instance, let d be a Boolean datatype where the lexical space = {1,0} (i.e. the list of legal values to represent the data type) and the 21st Computer Science Seminar SE1-T4-2
  • 3. value space = {True, False}. In this case, the lexical-to-value mapping maps the string 1 to True and the string 0 to False. In order to provide adequate support for value comparison, it is necessary for the query language to exploit any data types inherent in the RDF model. • Closure: In the case of RDF, closure means that the results of any query operation are also graphs and not any other data structure. This means the results of the query can be used as input to another query and is useful when one wants to break a larger query into smaller queries[3]. More Advanced Features: • Semantic Capabilties: One of the main advantages of the RDF data model is that it provides built-in support for semantic inference via RDFS. Semantic support of a query language might include: • The ability to query the subClass/SubProperties of the result set • The ability to query the subClass or SubProperties of the schema • The ability for a query language to support multiple schemas due to inability to force content providers to use just one schema [12] • Optional Values: Most RDF-Query languages provide path expression support for exact matching on a graph. Some languages such as SPARQL provide support for specifying optional sections of the graph similar to an outer join in the world of SQL. With this functionality, results can be returned based on partial matches of pattern data. • Aggregate Functions: Aggregate functions such as MIN, MAX and COUNT have long been supported in the structured world of SQL and could provide benefit to the RDF community. However, very few current languages support aggregation. •Advanced Set Operations: Within the relational database model, several algebraic operations are implemented which can be combined . These 5 operations form the basis of relational completeness and include selection, projection, Cartesian product, set difference and set union [4]. The three basic operations selection, projection and product are supported by all of the languages presented in this paper via the path expression; however, some languages also support set union and set intersection [4]. 3. QUERY LANGUAGE SURVEY 3.1 Overview A survey of popular RDF query languages recently conducted by the W3C identified more than 20 languages that are either under development or have been implemented [6]. Some of the languages developed provide a basic level of pattern matching with no inference capabilities while others allow for more complex semantic processing. Still others were created as extensions of existing technologies languages such as X-Query to provide perhaps an easier transition for users of XML-based query languages. This paper will not perform a complete survey of all RDF-based query languages, but provide a sampling of languages that represent a broad cross-section of approaches including: • A language with major mainstream industry support (RDQL) which gives insight into the level of current commitment within mainstream technology companies such as Hewlett Packard • A language that has strong W3C support (SPARQL) which gives insight into the future direction of standardization efforts within the W3C • A language that leverages current XML approaches (XsRQL) which gives insight into how RDF-Query can be added as an extension to other technologies • A language with strong collaboration between the open source community and industry (SeRQL) and gives insight into an alternative approach to W3C standardization In order to perform a simple side-by-side comparison of languages listed below, this paper will note whether each query language provides basic support for the 7 language features aforementioned. Whereupon a language offers advanced support for a feature listed below, it will be noted. 1. Value comparison and data type support 2. Generalized path expressions 3. Closure 4. Semantic capabilities 5. Optional values 21st Computer Science Seminar SE1-T4-3
  • 4. 6. Aggregate functions 7. Advanced set operations (union, intersection) 3.2 RDQL RDQL was developed by Hewlett Packard and submitted to the W3C in January 2004. It has been implemented in a several RDF systems including Jena, RDFStore, Sesame, PHP XML Classes, 3 Store and RAP-RDF API for PHP [13]. RDQL was derived mainly from SQUISH, an earlier language. The syntax of RDQL is similar to a SQL-like select pattern where the select clause allows the projection of the variables [4]. Based on the limited amount of functionality offered in RDQL, it would appear that RDQL was designed to be a simple language. The RDQL approach “suggests a strategy for possible standardization: an RDQL- like language could be developed and deployed without detailed treatment of rule or inference facilities, yet subsequently we used to query ‘smarter’ RDF services which make use of inferences licensed by OWL or RDF-rule semantics” [15]. Syntax Example: A simple path expression SELECT ? x WHERE (?x, http://www.w3.org/1999/02/22-rdf- syntax,ns#type,http://example.com/someType>) This select statement matches all statements in the graph that have a property of http://www.w3.org/1 999/02/22-rdf- syntax-ns#type and object of http://example.com/someType. The variable ?x is bound to the subject resource [13]. Table 1: RDQL Scorecard Data Type Supp. Path Exp. Closure Seman Funct. Opt Values Agg Funct. Adv Set Ops Yes X X No X X X X X 3.3 SPARQL SPARQL has been designed to meet the requirements and design goals as described in the W3C RDF Data Access Working Group (DAWG) document “RDF Data Access Use Cases and Requirements” [12]. To date, SPARQL has not been endorsed by the working group and there have been a number of design issues raised to the working group. Although it is an incomplete implementation, it represents the latest work of the W3C and provides insight into the future direction of standardization efforts within the W3C. It would appear from the SPARQL Working Draft that the language developers are following an iterative development process vis-à-vis incorporating best practices from other languages as well as direct input from the W3C RDF Data Access Working Group. As such, it is difficult to say how much additional functionality SPARQL will have when it is finally complete. SPARQL offers many of the basic features desired in an RDF- based query language. SPARQL provides a subset of operations on plain literals, XSD integers and XSD floats defined in XQuery and XPath functions and operators such as comparison of numeric values, functions on string values casting, comparison of duration, time and date values [12]. SPARQL treats an RDF graph as pure data and is not aware of any inference that an rdf store may provide [12]. One interesting feature of SPARQL is the RDF keyword CONSTRUCT where query results are returned as an RDF graph specified by an rdf template. Specifically, the CONSTRUCT keyword returns a single RDF graph formed as the union of the graph template with variable values obtained from each query result [12]. Syntax Example: An example of how SPARQL implements Optional matching PREFIX foaf: http://xmlns.com/foaf/0.1/ SELECT ?name ?mbox WHERE (?x foaf:name ? name) OPTIONAL (?x foaf:mbox ?mbox) [12] This query finds the names of people in the dataset and if there is a mbox property, it will return that as well. The main power of the OPTIONAL statement is that allows a partial match on the target graph similar to an outer join in SQL. Without the OPTIONAL keyword, each triple pattern in a WHERE clause must be satisfied because the triple patterns are essentially ‘anded’ together to enable the binding of variables to values. Table 2: SPARQL Scorecard Data Type Supp. Path Exp. Closure Seman Funct. Opt Values Agg Funct. Adv Set Ops Yes X X X X No X X X 3.3 SeRQL SeRQL (pronounced “circle”) is loosely based on several earlier languages such as RQL, RDQL and N3 [4] and as such represents a second generation language. SeRQL’s aim is to reconcile ideas from existing proposals into a proposal that satisfies a lot of these key requirements [3]. The overall design goal of SeRQL includes compiling and implementing many of the best features from earlier languages and delivering a light-weight yet powerful language [4]. One interesting characteristic of SeRQL is that it 21st Computer Science Seminar SE1-T4-4
  • 5. represents a collaboration between industry and the open source community and seemingly offers a viable alternative to RDF- query standardization over the W3C process. It is being developed as a collaboration between the Dutch software company Aduna and the open source community to address many of the concerns of the Sesame user community (www.openrdf.org) [3]. Sesame is an open source RDF database that offers support for many query languages including SeRQL which has become the default query language for Sesame. SeRQL supports many of the basic RDF-query features such as path expressions, Boolean constraints and advanced features such as optional matching. Advanced features such as functions for aggregation are not supported. Syntax Example: One of the more interesting features of SeRQL is its support for advanced path expressions. In situations where one wants to query =two or more triples with identical subject and predicate, the subject and predicate do not have to be repeated. SeRQL: [subj] pred1 [obj1, obj2, obj3] This expression is equivalent to [subj1] pred1[obj1] [subj1] pred1[obj2] [subj1] pred1[obj3] [3] SeRQL also supports branched path expressions, which is useful when multiple properties result from a single node SeRQL:[subj1] pred1[obj1] pred2[obj2] which is equivalent to: [subj1] pred1[obj1] [subj1] pred2[obj2] [3] Another interesting syntax example is how SeRQL supports RDF reification. Reification is where the subject or object of a statement is itself a statement. Since it is asyntactic construction, it can be expressed using basic path expression syntax. [3] [statemen1t] <rdf:type>[rdf:statement], [statement1] <rdf:subject>[subj1], [statement1] <rdf:predicate>[pred1], [statement1] <rdf:object>[obj1], [statement1] <rdf:pred2>[obj2], This is a cumbersome way of dealing with reification. SeRQL allows one to treat reified statements with a shorthand. SeRQL:{subj1} pred1 {obj1} pred2 {obj2} [3] The principles of navigating path expressions are also applied to navigating class and property hierarchies within SeRQL [3]. To retrieve the subclasses of a particular class:<my:class1> {subclass} <rdfs:subClassOf> <my:class1> [3] Table 3 SeRQL Scorecard Data Type Supp. Path Exp. Closure Sem Funct. Opt Values Agg Funct. Adv Set Ops Yes X X X X X No X X 3.4 XsRQL XsRQL is an RDF query language that derives most of syntax and style from X-Query while leveraging many of the useful features developed by the W3C XML Query Working Group while omitting parts of the X-Query specification that are specific to XML and hence not required in an RDF environment [8]. The interaction between query evaluation and the data model in XsRQL is similar to the X-Query: as expressions are evaluated within XsRQL, they populate instances into the result set from the target graph. [8]. As the query processor evaluates higher expressions on the query tree, new items populate the result set or existing items disappear and the whatever remains at the top of the query tree are the results of the query [8]. Because of the relative maturity of XQuery, XsRQL really shines with its support of built in functions and currently supports string functions such as chr, empty, exists, string-length as well as aggregate functions such as count [8]. XsRQL does not follow the familiar SQL syntax of the language examples previously mentioned in this paper, rather it follows the slightly abbreviated syntax of XQuery. Syntax Example Give me every vcard :Full name in the repository RDQL SQL-like syntax: select ?x, ?fname where (?x <http://www.w3.org/2001/vcard- rdf/3.0#fn> ?fname [8] In the abbreviated world of XsRQL, the query would look like this: @< http://www.w3.org/2001/vcard-rdf/3.0#fn>/* [8] Another example where we want to find within a datastore triples where the age of the first partner is less than that of the second partner In RDQL: 21st Computer Science Seminar SE1-T4-5
  • 6. SELECT ?x, ?y WHERE (?x:marriedTo ?y) ?(x :age ?xAge) (?y :age ?yAge) and ?xAge < ?yAge In XsRQL: for $x in * [@<marriedTo>] for $y in $x/@<marriedTo>/* return {$x,@marriedTo>, $y}[8] This example more fully illustrates how different the XsRQL language syntax is compared to other RDF query languages. Table 4 XsRQL Scorecard Data Type Supp Path Exp. Closure Sem Funct. Opt Values Agg Funct. Adv Set Ops Yes X X X No X X X X 4. RESULTS Table 5 : Combined Scorecard Data Type Sup Path Exp. Closur e Seman Funct. Opt Values Agg Funct. Ad v Set Ops SeRQL Yes Yes Yes Yes Yes No No SPARQ L Yes Yes Yes No Yes No No RDQL Yes Yes No No No No No XsRQL Yes Yes No No No Yes No 5. CONCLUSIONS From the basis of this scorecard, SeRQL provides the most language features to date with SPARQL coming in a close second. In addition, both SeRQL and SPARQL provide an easy to read SQL-like syntax. It would appear that SeRQL with strong support from the open source community and SPARQL with strong support from the W3C are leading the charge towards standardization. As previously stated, standardization for RDF Query is an important step in the evolution of the Semantic Web as it will likely foster the necessary widespread industry commitment to make the Semantic Web a success. It will be interesting to see how quickly the W3C can forward its standardization efforts with the open source community progressing as fast as it is. Languages such as XsRQL will likely struggle to gain a wide foothold due to the arcane syntax of the language. However, it is possible that one size may not fit all and one language may not be enough to support the diverse needs of the RDF community. Some of the challenges to widespread adoption of RDF Query are tightly coupled with the challenges of the Semantic Web. As of January 2005, Peter Norvig, Director of Google’s Search Quality, reports that there are less than 200,000 documents on the web that have and .rdf or .owl extension which is approximately .0005% of the web [16]. What is the point of building RDF query tools if there are no documents to query? [16]. Another challenge to the Semantic Web and consequently RDF Query is competing ontology’s or vocabularies [16]. Content providers within a well- defined vertical industry might be successful defining ontology’s based on earlier successes with ANSI X.12 which defines industry specific EDI transaction sets because customers often force suppliers to comply [16]. However, it will be difficult to enforce ontology’s within a broader pool of content providers. Lastly, RDF Query to provide value, the metadata people affix to content must be trustworthy and in some cases content providers may be motivated to provide inaccurate or misleading metadata to lure search users. In summary, their is a symbiotic relationship between RDF Query and the larger Semantic Web initiative and it will be important for RDF Query standardization efforts to align with the larger issues that the Semantic Web initiative is encountering in order to gain widespread adoption. 6. REFERENCES [1] S. Alexaki, V. Christophides, G. Karvounrakis and D. Plexousakis, “RQL: A Declarative Query Language for RDF”, Institute of Computer Science, Greece”,[online document] available at http://www2002.org/CDROM/refereed/329/ [2] J. Angele, D Brickley, S. Decker and J. Saarela, “A Query and Inference Service for RDF”, January 1998 ”,[online document] http://www.w3.org/TandS/QL/QL98/pp/queryservice.html [3] J. Broekstra, A. Kampman, “Se-RQL: An RDF Query and Transformation Language, Draft”,[online document] available at http://www.cs.vu.nl/~jbroeks/papers/SeRQL.pdf [4] J. Broekstra, A. Eberhart, P. Haase, and R. Volx, “A Comparison of RDF Query Languages”, Proceedings of the Third International Semantic Web Conference, Hiroshima, Japan, November 2004”,[online document] available at http://www.cs.vu.nl/~jbroeks/papers/survey.pdf [5] K Clark, RDF Data Access Use Cases and Requirements, October 2004”,[online document] available at http://www.w3.org/TR/rdf-dawg-uc/ [6] B. Grosof, E. Prud'hommeaux, RDF Query Survey, [online document], Available at http://www.w3.org/2001/11/13-rdf- query-rules/ 21st Computer Science Seminar SE1-T4-6
  • 7. [7] G. Karvounrakis, “RDF Query Languages: A state-of-the-art”, Institute of Computer Science, Greece online document] available at http://139.91.183.30:9090/RDF/publications/state.html [8] H. Katz, “XsRQL:an X-Query-style Query Language for RDF, A Submission to the RDF Data Access Working Group” [online document], June 2004, Available at http://www.fatdog.com/xsrql.html [9] F. Manola, B.McBride, and E. Miller, Eds., RDF Primer , W3C Recomendation ,February 2004 ”,[online document] available at http://www.w3.org/TR/rdf-primer/ [10] L. Miller 1 , A. Seaborne 2 , A. Reggiori 3 ,” Three Implementations of SquishQL, a Simple RDF Query Language”, 1 ILRT, Bristol University, UK, 2,3 Hewlett-Packard Laboratories, [online document], available at http://ilrt.org/discovery/2002/05/squish- iscw/http://ilrt.org/discovery/2002/05/squish-iscw/ [11] W. Nejdl, W. Siberski and M. Sintek, “Design Issues and Challenges for RDF and Schema–Based Peer-to-Peer Systems”, SIGMOD Record, Vol 32, No. 3, September 2003, pp. 41- 46 [12] E. Prud'hommeaux, A. Seaborne, “ SPARQL Query Language for RDF”, W3C Working Draft 12 October 2, [online document], Available at http://www.w3.org/tr/rdf-sparql-query/ [13] A. Seaborne, “RDQL, A Query Language for RDF,W3C Member Submission, [online document], January 2004, Available at http://www.w3.org/submission/2004/subm-rdql-20040109 [14] W3C, RDF Query and Rules”,[online document] available at http://www.w3.org/2001/11/13-RDF-Query-Rules/terms [15] XML-Cover-Pages.Org, Hewlett-Packard Submits Query Language for RDF (RDQL) to W3C, [online document], January 2004, Available at http://www.xmlcoverpages.org/ni2004-01-14- a.html [16] P. Norvig, Semantic,”Web Ontologies: What Works and What Doesn't”,online document, January 2005, available http://www.alwayson- network.com/comments.php?id=7480_0_3_0_C 21st Computer Science Seminar SE1-T4-7