=Paper=
{{Paper
|id=Vol-1215/paper-06
|storemode=property
|title=Validating and Describing Linked Data Portals using RDF Shape Expressions
|pdfUrl=https://ceur-ws.org/Vol-1215/paper-06.pdf
|volume=Vol-1215
|dblpUrl=https://dblp.org/rec/conf/i-semantics/GayoPSR14
}}
==Validating and Describing Linked Data Portals using RDF Shape Expressions==
https://ceur-ws.org/Vol-1215/paper-06.pdf
Validating and Describing Linked Data Portals using RDF
Shape Expressions
Jose Emilio Labra Gayo Eric Prud’hommeaux
University of Oviedo World Wide Web Consortium (W3C) MIT,
Dept. of Computer Science Cambridge, MA, USA
C/Calvo Sotelo, S/N eric@w3.org
labra@uniovi.es
Harold Solbrig Jose María Álvarez Rodríguez
Mayo Clinic Dept. Computer Science
College of Medicine, Rochester, MN, USA Carlos III University
josemaria.alvarez@uc3m.es
ABSTRACT technology or as an interoperability layer. However, there is a lack
In order to improve the quality of linked data portals, it is necessary of an accepted practice to declare and constrain the shape of an
to have a tool that can automatically describe and validate the RDF RDF graph in a way that can be automatically validated to augment
triples exposed. the quality of RDF based data portals.
RDF Shape Expressions have been proposed as a language based Validation is standard practice in other conventional data languages.
on Regular Expressions that can describe and validate the structure Industrial setting count on parsing grammars for domain-specific
of RDF graphs. languages, DDL constraints in SQL databases, and W3C XML
Schema or RelaxNG for XML documents.
In this paper we describe the WebIndex, a medium sized linked
data portal, and how we have employed Shape Expressions to doc- In the case of RDF, although there are standards for inference like
ument its contents and to automatically validate the shapes of the RDF Schema and OWL, these technologies employ Open World
resources. and Non-Unique Name Assumptions that create difficulties for val-
idation purposes [18].
Categories and Subject Descriptors
I.2.4 [Knowledge Representation Formalisms and Methods]: Rep- The RDF Shape Expressions language is intended to perform the
resentation languages; H.3.5 [Online Information Services]: Web- same function for RDF graphs as Schema languages to XML. It
based services can be used to validate documents, communicate expected graph
patterns for interfaces, and generate user interface forms and code.
General Terms The syntax and semantics of Shape Expressions are designed to be
Theory familiar to users of regular expressions (specially RelaxNG). The
main difference is that RDF data is a set (of triples) while regular
Keywords expression data is a sequence (of characters). Regular expressions
RDF, Graphs, Validation, Transformation correlate an ordered pattern of atomic characters and logical opera-
tors against an ordered sequence of characters while Shape Expres-
1. INTRODUCTION sions correlate an ordered pattern of pairs of predicate and object
Linked Data portals have emerged as a way to publish data on the classes and logical operators against an unordered set of arcs in a
Web following a set of principles [1] which improve data reuse and graph.
integration. As indicated in [2], linked data relies on documents
using RDF representations to make typed statements that link ar- In this paper we propose the use of RDF Shape Expressions to de-
bitrary things in the world. RDF appears as a data integration lan- scribe the contents of Linked Data portals in a way that can be
guage and some linked data applications use RDF as a database automatically validated.
As a use case, we will describe the development of the 2013 We-
bIndex data portal1 , which is a linked data portal of medium size
(around 3.5 million of triples) that contains information about the
statistical computations that have been carried on to generate the
WebIndex. We have selected this use case because it contains a data
model with interrelated shapes and reuses several existing vocab-
Copyright is held by the author/owner(s).
LDQ 2014, 1st Workshop on Linked Data Quality Sept. 2, 2014, Leipzig, 1
Germany. http://data.webfoundation.org/webindex/
v2013
ularies like RDF Data Cube, Organization Ontology, Dublin Core,
etc. dataset:DITU a qb:DataSet ;
rdfs:label "ITU Dataset " ;
dct:publisher org:ITU ;
2. WEBINDEX DATA MODEL qb:slice slice:ITU09B ,
The WebIndex is a multi-dimensional measure of the World Wide slice:ITU10B,
Web’s contribution to development and human rights globally. It ...;
covers 81 countries and incorporates indicators that assess several ...
areas like universal access; freedom and openness; relevant con- slice:ITU09B a qb:Slice ;
tent; and empowerment2 . qb:sliceStructure wf:sliceByArea ;
qb:observation obs:obs8165,
The 2012 version offered a data portal where the data was obtained obs:obs8166,
by transforming raw observations and precomputed values from ...
Excel sheets to RDF. The 2013 version of the WebIndex data portal ...
employs a new validation and computation approach that tries to org:ITU a org:Organization ;
obtain a verifiable linked data version of the Web Index data. rdfs:label "ITU" ;
foaf:homepage
The WebIndex data model is based on the RDF Data Cube vocab- .
ulary. Figure 1 represents the main concepts of the data model3 . country:Spain a wf:Country ;
wf:iso2 "ES" ; wf:iso3 "ESP" ;
As can be seen, the main concept are observations of type qb: rdfs:label " Spain "
Observation which have a float value cex:value and are re- .
lated to a country, a year a dataset and an indicator. indicator:ITU_B a wf:SecondaryIndicator ;
rdfs:label " Broadband subscribers %"
A dataset contains a number of slices, each of which also contains .
a number of observations.
Indicators are provided by an organization of type org:OrganizationA validator of this model can validate the simple structure of each
which employs the Organization ontology[15]. Datasets are also type of resource but it would be better if it can declare and detect
published by organizations. all these interrelationships.
As a sample of some data, an observation can be that Spain has The WebIndex data model also includes a linked data representa-
value 23.78 in 2011 for the indicator ITU-B (Broadband subscribers tion of computations so it is possible to declare the data from which
per 100 population) in the dataset DITU provided by ITU (Interna- a value has been computed so it can be checked. We omit those
tional Telecommunication Union). This information can be repre- classes for brevity.
sented in RDF using Turtle syntax as4 :
obs:obs8165 a qb:Observation ; 3. USING SHAPE EXPRESSIONS TO DE-
rdfs:label "ITU B in ESP , 2011" ; SCRIBE THE WEBINDEX DATA MODEL
dct:issued In this section we will describe the WebIndex data model using
"2013 -05 -30 T09 :15:00 " ^^xsd:dateTime ; Shape Expressions.
cex:indicator indicator:ITU_B ;
qb:dataSet dataset:DITU ; The RDF Shape Expressions language is inspired by RelaxNG and
cex:value 23.78^^xsd:float ; also has two syntaxes: a compact one and an RDF serialization.
cex:ref-area country:Spain ; The compact syntax is more oriented towards human readability
cex:ref-year 2011 ; while the RDF serialization can be employed to exchange and store
...other properties omitted for brevity shape expressions using standard semantic web tools. A primer to
. the Shape Expressions language can be found at http://www.
w3.org/2013/ShEx/Primer.
Notice that the WebIndex data model contains data that is com- A shape expression is a labelled pattern for a set of RDF Triples
pletely interrelated. Observations are linked to indicators and datasets. sharing a common subject. Syntactically, it is a pairing of a label,
Datasets contain also links to slices and slices have links to indica- which can be an IRI or a blank node, and a rule enclosed in brackets
tors and observations again. Both datasets and indicators are linked ({ }). Typically, this rule is a conjunction of constraints separated
to the organizations that publish or provide them. by commas (,). For example, we can declare the shape of a country
as:
The following example contains a sample of interrelated data for
this domain. {
2
a (wf:Country)
http://thewebindex.org , rdfs:label xsd:string
3
In the paper we will employ common prefixes that can be found , wf:iso2 xsd:string
in http://prefix.cc
4 , wf:iso3 xsd:string
The URI of the real observation is http://
data.webfoundation.org/webindex/v2013/ }
observation/obs8165
Figure 1: Simplified WebIndex data model
The above declaration indicates that a country must have rdf: }
type with value wf:Country. It must also have the properties
rdfs:label, wf:iso2 and wf:iso3 with a value of type xsd
:string. The declarations for observations and indicators are similar:
The semantics of Shape Expression validation acts as a type infer- {
ence system which infers a type (shape) for a given node in an RDF a (qb:Observation)
graph. , cex:value xsd:float
, dct:issued xsd:dateTime
With the previous declaration, a Shape Expressions validator would , rdfs:label xsd:string?
infer: , qb:dataSet @
, cex:ref-area @
country:Spain , cex:indicator @
, cex:ref-year xsd:gYear
The Shape Expressions language is inspired by Regular Expres- }
sions and the rules can contain cardinality constraints with the val-
ues + (one or more), * (zero of more), ? (zero or one) and even {
ranges {m,n} (between m and n repetitions). a ( wf:PrimaryIndicator
wf:SecondaryIndicator
It is also possible to declare that the value of some property has a )
given shape using the @ character. , rdfs:label xsd:string
, rdfs:comment xsd:string ?
For example, the shape of datasets can be described as: , skos:notation xsd:string ?
}
{
a (qb:DataSet)
, qb:structure (wf:DSD)
Finally, organizations can be declared as:
, rdfs:label xsd:string?
, qb:slice @+ {
} a ( org:Organization )
, rdfs:label xsd:string
, foaf:homepage IRI
which declares that a dataset must have rdf:type with value qb: , org:hasSubOrganization
DataSet and qb:structure with value wf:DSD. It may have @
a rdfs:label with a value of type xsd:string and must have }
one or more slices with the shape Slice.
In a similar way, it is possible to declare slices as: As can be seen, Shape Expressions offer an intuitive way to de-
scribe the contents of linked data portals. In fact, we have em-
{ ployed Shape Expressions to document both the WebIndex5 and
a (qb:Slice) Landbook6 data portals. The documentation defines templates for
, qb:sliceStructure ( wf:sliceByYear )
5
, qb:observation @+ http://weso.github.io/wiDoc
6
, cex:indicator @ http://weso.github.io/landportalDoc/data
the different shapes of resources and for the triples that can be re- • By dereference: The RDF triples are obtained by dereferenc-
trieved when dereferencing those resources. ing the URIs of the resources that will be validated and using
content negotiation to ask for RDF/XML or Turtle represen-
These templates define the dataset structure in an intuitive way and tations.
can be used to act as a contract between developers of the data por-
tal. We noted that having a good data model with its corresponding
Shape Expressions specification facilitated the communication be- The RDFShape tool allows the user to specify whether to use a
tween the different teams involved in the development of the data Shape Expression schema or not. If not, the tool just checks that
portal. the RDF can be parsed. Otherwise, the user can also enter a Schema
by URI, by File or by Input.
4. IMPLEMENTATIONS OF SHAPE EXPRES-
Finally, it is possible to validate a specific IRI or just any IRI in the
SIONS RDF graph. Specifying an IRI is recommended when validating by
Currently, there are four implementations of Shape Expressions in Endpoint to check the shape of a given IRI in the endpoint and it
progress: is mandatory when using by dereference, as it will be the IRI that
will be dereferenced to validate its representation.
• FancyShExDemo7 was the first prototype implementation in
Figure 2 contains a screen capture of the RDFShape validation tool.
Javascript. It handles semantic actions which can be used to
extend the semantics of shape expressions and even to trans-
form RDF to XML or JSON. It supports a form-based sys- 6. VALIDATING LINKED DATA PORTALS
tem with dynamic validation during the edition process and USING SHAPE EXPRESSIONS
SPARQL queries generation. RDF Shape Expressions can be used not only to describe the con-
8
• JSShexTest , developed by Jesse van Dam is another Javascript tents of linked data portals, but also to validate them.
implementation. It both supports the SHEXc and SHEX/RDF
syntax of Shape Expressions and contains a validation se- We consider that one of the first steps in the development of a linked
mantics for testing purposes based on truth tables. data portal should be the Shape Expression declarations of the dif-
ferent types of resources. Shape Expressions can play a similar role
• Shexcala9 : an implementation developed in Scala with an to Schema declarations in XML based developments. They can act
efficient implementation based on derivatives of regular ex- as a contract for both the producers and consumers of linked data
pressions. It supports validation against an RDF file and portals.
against a SPARQL endpoint. In the following section we
describe an online validation service which is implemented Notice, however, that this contract does not suppose an extra-limitation
on top of ShExcala. between the possible consumers a linked data portal can have. There
is no impediment to have more than one shape expressions which
• Haws10 : a Haskell implementation based on type inference enforce different constraints. As a naïve example, the declarations
semantics and backtracking. This implementation can be of the iso2 and iso3 code of Countries can be further constrained
seen as an executable monadic semantics of Shape Expres- using regular expressions to indicate that they must be 2 or 3 alpha-
sions [10]. betical characters or could be more relaxed saying that it may be
any value (not only strings). The advantage of Shape Expressions
5. RDFSHAPE: AN RDF SHAPE VALIDA- is that they offer a declarative and intuitive language to express and
TION SERVICE refer to those constraints.
RDFShape11 is an online RDF Shape validation web service that
can be used to validate both the syntax and the shape of RDF data Shape Expression declarations can also be employed to generate
against some schema. synthetic linked data in the development phase so one can perform
stress tests. For example, during the development of the WebIndex
The online service has five types of inputs for RDF: data portal, we implemented the wiGenerator12 tool which is a sim-
ple program that can generate random linked data that follows the
WebIndex data model with any number of indicators, years of coun-
• By URI: The RDF data to be validated is downloaded from a tries specified by the user. These fake RDF datasets can be em-
given URI ployed to perform stress and usability tests of the data visualization
software.
• By File: The data is uploaded from a local file
Shexcala offers the possibility to validate a URI in an endpoint or
• By Input: The data is inserted in a textarea
by dereferencing it (retrieving the RDF data behind that URI). The
• By Endpoint: The RDF data triples are retrieved from a SPARQL implementation performs a generic SPARQL query to obtain all the
endpoint on demand. The user has to provide the URI of the triples that have a given node as subject in the endpoint:
endpoint.
construct { $node ?p ?y } where {
7
http://www.w3.org/2013/ShEx/FancyShExDemo $node ?p ?y .
8
https://github.com/jessevdam/shextest }
9
http://labra.github.io/ShExcala/
10
http://labra.github.io/haws/
11 12
http://rdfshape.weso.es http://labra.github.io/wiGenerator/
Figure 2: Screen capture of RDFShape tool
Once the triples are retrieved, the system validates the Shape Ex- data portals, nothing precludes to define templates and libraries of
pressions declarations of that graph to check the shape of that node. generic shapes that can be reused between different data portals.
In this way, it is very easy to perform shape checking on the con- 7. EXTENSIONS AND CHALLENGES
tents of linked data portals. For example, one can retrieve all the At the moment of this writing, the W3C has just chartered a RDF
nodes of type qb:Observation and check that they have the Data Shapes Working Group with the mission to produce a lan-
shape . guage for defining structural constraints on RDF graphs. The Shape
Expressions language is being used as part of the working group
Notice that in general, this kind of validation is context sensitive to discussions so it is possible that some parts of the language will
a given data portal. Shape Expressions deliberately separates types change in the future.
from shapes.
There are currently several topics and extension proposals for the
For example, LandPortal also expresses a shape for its use of qb: Shape Expression language that may be interesting to mention:
Observation. Both WebIndex and LandPortal respect the RDF
data Cube definition of an Observation, but they can require or pro-
hibit different properties (from that ontology or elsewhere) on those • The Shape Expression language contains other common reg-
Observations. The observations in WebIndex have different shapes ular expression operators like alternatives (|), negations (!),
than the observations in LandPortal, but all of them have type qb groupings using parenthesis, etc. that can express more com-
:Observation without introducing any logical conflicts. This plex patterns. For example, we could declare that Countries
reflects a different usage pattern than generally seen for OWL or can have either wf:iso2 or wf:iso3, and that they must
SPIN constraints (see 8). We consider that this difference between not have the property dc:creator as:
structural shapes and semantic types of resources improves the sep-
aration of concerns involved in linked data portal development. { a (wf:Country)
, rdfs:label xsd:string
Nevertheless, although some shapes can be specific to some linked , ( wf:iso2 xsd:string
| wf:iso3 xsd:string The Javascript implementation supports semantic actions in
) Javascript and SPARQL which can add more expressiveness
, ! dc:creator . to the validation declarations. In fact, it also contains two
} simple languages (GenX and GenJ) which enable an easy
way to transform RDF to both XML and JSon.
• Open vs Closed shapes. An open shape is a shape expression Following the RelaxNG path, the Shape Expression language
that validates nodes that contain the triples specified in the can be seen as a simple Domain Specific Language which
shape but can also contain other triples. A closed shape only is tailored to express the structure of RDF graphs. It is not
validates nodes with those triples and no more. For example, intended to perform strong constraint checking or validation
if we declare users shapes as: using computations. However, with semantic actions or shape
expression validators embedded in other tool chains that pos-
{ a foaf:Person } sibility could be offered.
In the same way, the interplay between Shape Expressions
and we have the following triples: and reasoners is not established. Some applications could do
inference between checking the shape of RDF graphs, while
:john a foaf:Person,
other applications could check the shapes before invoking a
foaf:name "John" .
reasoner. Another possibility that could be explored is to
have some built-in way that could invoke reasoning capabil-
Using closed shapes, the system would not assign :john ities.
the shape because it contains an extra triple, while
using open shapes it would assign it that shape.
It is perceived that open shapes fit better in an Open World One of the challenges of the Shape Expressions language is the
web, while closed shapes would be better for more controlled performance of Shape checking. A naïve implementation of Shape
environments. checking using backtracking can lead to exponential growth. We
have found that using regular expression derivatives offers an ef-
In this line of work, there is also a proposal to reuse shape
ficient implementation and we are currently evaluating its perfor-
descriptions by including other shape declarations. For ex-
mance.
ample, one may be interested to say that providers have the
shape but also contain the property wf
:sourceURI as: 8. RELATED WORK
Improving the quality of linked data has been of increasing interest
& in the last years. Sieve[11] proposed a framework for expressing
{ wf:sourceURI IRI } quality assessment methods as well as fusion methods. RDFU-
nit[8] is a test-driven framework that can run test cases against an
• Incoming edges, relations and named graphs. The current endpoint. In the case of RDF validation, the main approaches can
Shape Expression language is based on describing the sub- be summarized as:
jects of an RDF graph. It would be possible to extend the
language to handle also objects and properties. For example,
• Inference based approaches, which try to adapt RDF Schema
we can declare reverse arcs using the operator ^ to indicate
or OWL to express validation semantics. The use of Open
incoming arcs. The Country declaration could be:
World and Non-unique name assumption limits the valida-
{ a (wf:Country) tion possibilities. In fact, what triggers constraint violations
, rdfs:label xsd:string in closed world systems leads to new inferences in standard
, wf:iso2 xsd:string OWL systems. [4, 18, 12] propose the use of OWL expres-
, wf:iso3 xsd:string sions with a Closed World Assumption to express integrity
, ^ cex:ref-area @ * constraints.
}
• SPARQL Inferencing Notation (SPIN)[7] constraints asso-
ciate RDF types or nodes with validation rules. These rules
with the meaning that a country can receive (zero or more) are expressed as SPARQL ASK queries where true indi-
arcs with property cex:ref-area of shape . :ConstraintViolations. SPIN constraints use the ex-
In the same way, it may be interesting to declare the shape pressiveness of SPARQL plus the semantics of the ?this
of RDF nodes that act as properties. Another extension pro- variable standing for the current subject and the spin:
posal is to describe named graphs. These two proposals are ConstraintViolation class.
not difficult to add, but it is not clear which syntax would be
intuitive enough for them. • SPARQL-based approaches use the SPARQL Query Lan-
gugage to express the validation constraints. SPARQL has
• More expressiveness. The Shape expression language can much more expressiveness than Shape Expressions and can
be extended with semantic actions to increase the expres- even be used to validate numerical and statistical compu-
siveness of the language. Semantic actions are marked by tations [9]. However, we consider that the Shape Expres-
%lang { actions %} which means that the validator sions language will be more usable by people familiar with
can invoke a processor of the language lang with the corre- validation languages like RelaxNG. Nevertheless, Shape Ex-
sponding actions. pressions can be translated to SPARQL queries. In fact, we
have implemented a translator from Shape Expressions to The complexity of the validation algorithms for Shape Expressions
SPARQL queries. This translator combined with semantic offers some theoretical challenges related to regular bag expres-
actions expressed in SPARQL can offer the same expressive- sions that have been tackled in [3]. The last implementation of
ness as other SPARQL approaches with a more succinct and Shexcala contained an algorithm based on derivatives of regular
intuitive syntax. expressions which greatly improved the efficiency of the validation
process.
There have been other proposals using SPARQL combined
with other technologies. Fürber and Hepp[6] proposed a
Although the language is new and the syntax can seem strange at
combination between SPARQL and SPIN as a semantic data
first sight, we noticed that people are able to learn the syntax and
quality framework, Simister and Brickley[17] propose a com-
to declare shape expressions quickly.
bination between SPARQL queries and property paths which
is used in Google and Kontokostas et al [8] proposed RDFU-
In general we consider that the benefits of validation can help the
nit a Test-driven framework which employs SPARQL query
adoption of RDF based solutions where the quality of data is an
templates that are instantiated into concrete quality test queries.
important issue.
We consider that Shape Expressions can also be employed in
the same scenarios as SPARQL while the specialized valida-
tion nature of Shape Expressions can lead to more efficient 10. REFERENCES
implementations. [1] T. Berners-Lee. Linked-data design issues. W3C design issue
document, June 2006.
• Grammar based approaches define a domain specific lan- http://www.w3.org/DesignIssue/LinkedData.html.
guage to declare the validation rules. OSLC Resource Shapes [16] [2] C. Bizer, T. Heath, and T. Berners-Lee. Linked data - the
have been proposed as a high level and declarative descrip- story so far. International Journal Semantic Web Information
tion of the expected contents of an RDF graph expressing Systems, 5(3):1–22, 2009.
constraints on RDF terms. Shape Expressions have been in- [3] I. Boneva, J. E. Labra, S. Hym, E. G. Prud’hommeau,
spired by OSLC although they offer more expressive power. H. Solbrig, and S. Staworko. Validating RDF with Shape
Dublin Core Application Profiles [5] also define a set of val- Expressions. ArXiv e-prints, Apr. 2014.
idation constraints using Description Templates with less ex- [4] K. Clark and E. Sirin. On RDF validation, stardog ICV, and
pressiveness than Shape Expressions. assorted remarks. In RDF Validation Workshop. Practical
Assurances for Quality RDF Data, Cambridge, Ma, Boston,
September 2013. W3c,
The main inspiration for Shape Expressions has been RelaxNG [19], http://www.w3.org/2012/12/rdf-val.
a Schema language for XML that offers a good trade-off between [5] K. Coyle and T. Baker. Dublin core application profiles.
expressiveness and validation efficiency. The semantics of Re- separating validation from semantics. In RDF Validation
laxNG has also been expressed using inference rules in the spec- Workshop. Practical Assurances for Quality RDF Data,
ification document [14] and is based on tree grammars [13]. In the Cambridge, Ma, Boston, September 2013. W3c,
case of Shape Expressions the underlying semantics can be defined http://www.w3.org/2012/12/rdf-val.
in terms of regular bag expressions [3]. [6] C. Fürber and M. Hepp. Using sparql and spin for data
quality management on the semantic web. In
Shape Expressions are also being employed in the development of W. Abramowicz and R. Tolksdorf, editors, Business
more specialized validators. For example, the Vaskos project13 is Information Systems, volume 47 of Lecture Notes in Business
developing a SKOS validator using a combination between Shape Information Processing, pages 35–46. Springer, 2010.
Expressions and SPARQL queries. [7] H. Knublauch. SPIN - Modeling Vocabulary. http:
//www.w3.org/Submission/spin-modeling/,
2011.
9. CONCLUSIONS
[8] D. Kontokostas, P. Westphal, S. Auer, S. Hellmann,
Shape Expressions have been proposed as a Domain Specific Lan-
J. Lehmann, R. Cornelissen, and A. Zaveri. Test-driven
guage that can describe and automatically validate RDF. They of-
evaluation of linked data quality. In Proceedings of the 23rd
fer a more expressive way to define sets of RDF graph shapes than
International Conference on World Wide Web, WWW ’14,
OSLC’s Resource Shapes or Dublin Core’s Application Profiles.
pages 747–758, Republic and Canton of Geneva,
There are trade-offs between expressiveness and implementability,
Switzerland, 2014. International World Wide Web
but compared to schema languages in other data models, Shape Ex-
Conferences Steering Committee.
pressions represent a conservative point in that spectrum, emulating
mostly the expressiveness of RelaxNG. [9] J. E. Labra and J. M. Alvarez Rodríguez. Validating
statistical index data represented in RDF using SPARQL
From a tooling perspective, shape expressions can be used stand- queries. In RDF Validation Workshop. Practical Assurances
alone to validate RDF graphs and endpoints offering a dedicated for Quality RDF Data, Cambridge, Ma, Boston, September
language that can be implemented efficiently and generate special- 2013. W3c,
ized error messages for the concrete task of shape validation. http://www.w3.org/2012/12/rdf-val.
[10] J. E. Labra Gayo. Reusable semantic specifications of
Given that Shape Expressions can be translated to SPARQL queries, programming languages. In 6th Brazilian Symposium on
they can also be combined with other widely deployed infrastruc- Programming Languages, 2002.
ture. [11] P. N. Mendes, H. Mühleisen, and C. Bizer. Sieve: Linked
Data Quality Assessment and Fusion. In 2nd International
13
http://vaskos.chemaar.cloudbees.net/ Workshop on Linked Web Data Management (LWDM 2012)
at the 15th International Conference on Extending Database
Technology, EDBT 2012, March 2012.
[12] B. Motik, I. Horrocks, and U. Sattler. Adding Integrity
Constraints to OWL. In C. Golbreich, A. Kalyanpur, and
B. Parsia, editors, OWL: Experiences and Directions 2007
(OWLED 2007), Innsbruck, Austria, June 6–7 2007.
[13] M. Murata, D. Lee, M. Mani, and K. Kawaguchi. Taxonomy
of xml schema languages using formal language theory.
ACM Trans. Internet Technol., 5(4):660–704, Nov. 2005.
[14] OASIS Committee Specification. RELAX NG Specification:.
http://relaxng.org/spec-20011203.html, 2001.
[15] D. Reynolds. The Organization Ontology.
http://www.w3.org/TR/vocab-org/, 2014.
[16] A. G. Ryman, A. L. Hors, and S. Speicher. OSLC resource
shape: A language for defining constraints on linked data. In
C. Bizer, T. Heath, T. Berners-Lee, M. Hausenblas, and
S. Auer, editors, Linked data on the Web, volume 996 of
CEUR Workshop Proceedings. CEUR-WS.org, 2013.
[17] S. Simister and D. Brickley. Simple application-specific
constraints for rdf models. In RDF Validation Workshop.
Practical Assurances for Quality RDF Data, Cambridge,
Ma, Boston, September 2013. W3c,
http://www.w3.org/2012/12/rdf-val.
[18] J. Tao, E. Sirin, J. Bao, and D. L. McGuinness. Integrity
constraints in OWL. In Proceedings of the 24th AAAI
Conference on Artificial Intelligence (AAAI-10). AAAI,
2010.
[19] E. van der Vlist. Relax NG: A Simpler Schema Language for
XML. O’Reilly, Beijing, 2004.