RDF/XHTML: Ontology Editing in HTML
Kun Ji
and Lauri Carlson
Department of Modern Languages, University of Helsinki, Helsinki, Finland
Keywords: Ontology Engineering, Semantic Web, Ontology Sharing and Reuse, Natural Language Processing.
Abstract: Although global, the web and its standards are not language independent. Semantic Web document
standards remain skewed toward Western languages and alphabets. This causes problems for truly
multilingual terminology work in the Semantic Web. Some of the problems are discussed and remedied by
the proposals in this paper. In particular, we present a format for representing RDF triple sets in XHTML
and software that supports multilingual editing of RDF/OWL ontologies in this format. The format was
developed for and is used in TermFactory, an ontology based multilingual terminology management
environment.
1 INTRODUCTION
Resource Description Framework (RDF 2004) is a
modeling language originally meant for annotating
semantic meta data in the web. RDF is a graph based
data model that allows making statements about
resources (in particular Web resources, identified by
URIs) and their relationships in the form of subject-
predicate-object triples. RDF Schema (RDFS 2004)
adds vocabulary and axioms for defining classes and
their instances, and primitives for defining further
vocabulary. One such vocabulary is Web Ontology
Language (OWL 2009), a Web version of
description logic. OWL has a back translation to
RDF in which one OWL predicate or construct may
go over to a RDF graph composed of several triples.
Ontology languages like RDF and OWL have
become a significant format for describing complex
concept systems in areas such as natural sciences
and medicine. The main focus in ontology work has
been in the concept systems as such. Many large
scale domain ontologies are language neutral, in
practice, biased toward English. To the extent
multilingual or indeed natural language terminology
is included, it is provided as simple string labels.
Terms are usually not described as ontological
resources in their own right.
But nothing prevents describing natural language
terms as ontology resources as well (Buitelaar et al.
2011). TermFactory (Kudashev et al, 2010, Carlson,
2012) is an ontology-based terminology
management system that does just that. TermFactory
(TF) comprises an ontology schema, web API, and
platform for collaborative terminology work that is
based on explicit ontological representation of both
concepts and their designations.
2 TERMFACTORY
TermFactory (TF) is an architecture and a workflow
for distributed and collaborative terminology work
in a global multilingual context. TF applies
Semantic Web technologies to the representation of
specialised multilingual terms and related concepts.
It also provides a workflow by which terminologies
can be collected, updated and agreed about by
professionals in relevant fields all over the globe,
during their everyday work, using virtual work
platforms over the web.
TF can be considered a semantic web framework
for multilingual terminology work. It provides
ontology and terminology formats, format
conversions, query and edit tools, repositories, web
services. TF enables people to do professional
quality terminology work jointly or separately,
building their work on others’ efforts, maintaining
quality and consistency of the jointly developed
terminology.
With Semantic Web techniques TF aims to
achieve:
Openness and conformance: both conceptual
and linguistic content can be globally
identified and mechanically validated.
365
Ji K. and Carlson L..
RDF/XHTML: Ontology Editing in HTML.
DOI: 10.5220/0004138103650368
In Proceedings of the International Conference on Knowledge Engineering and Ontology Development (KEOD-2012), pages 365-368
ISBN: 978-989-8565-30-3
Copyright
c
2012 SCITEPRESS (Science and Technology Publications, Lda.)
Flexible reuse of content: different ontologies
and terminologies should be able to coexist,
complement one another, and co–develop on
separate sites.
Ease of implementation and deployment:
contents usable by third party tools, help
divide and conquer big ontologies.
For TermFactory web ontology based terminology
management, we found a need for editing tools for
non-ontologists to edit term ontologies that would be
simpler and more accessible to terminologists than
mainstream ontology editors. We designed
RDF/XHTML and its Web API as an answer to this
need. As the format and tools are quite generic with
little that is specific to term ontologies in particular,
we propose to present them here to the ontology
developer community at large.
3 RDF FORMATS
RDF has several serialization formats (file formats)
which vary in the way in which resources and triples
are encoded. For historical reasons, XML is the
official syntax for RDF. For multilingual ontologies,
RDF/XML is not a good choice. A concrete flaw of
RDF/XML for multilingual (or non-Latin, in
general) ontologies is that there is no provision for
coding property names containing non-Latin
characters. The only RDF/XML representation for
property names is XML element name (QName),
which has a restricted character repertoire. Non-
Latin property names require lengthy and unreadable
character encodings. A simple expedient would be to
extend RDF/XML with property elements identified
by full URIs, e.g.
<ex:label>term</ex:label>
could be written as
<rdf:Property
rdf:about=http://example.com#label”>
term
</rdf:Property>
A simpler alternative to RDF/XML is Turtle
(2010), a textual format for RDF graphs close to the
triple format. Turtle is terse and human readable.
Yet it too has limitations. Resource names cannot be
abbreviated with prefixes if they contain Turtle
reserved characters. It would be as well for Turtle
not to reserve punctuation characters, since Turtle
punctuation is conventionally separated by
whitespace anyway.
4 ONTOLOGY EDITING
Syntax editing of ontology triples can yield
unexpected results. Deletion of facts in general
involves difficult problems of nonmonotone
reasoning or belief revision. The best one can do is
avoid redundancies by using some normal form.
A normal form is a unique choice among
equivalent representations. Reduction to normal
form by term rewriting is what many reasoners in
effect do. RDF/OWL databases are supposed to keep
graphs in a nonredundant form to support updates.
The standard serializations of RDF do not
provide a unique normal form. Textual normal forms
for RDF have been proposed (Carroll/Stickler 2004,
Dau 2006, Gutierrez et al. 2011). Semantic normal
forms for some description logics have been
proposed (Hitzler/Eberhart 2007, Bienvenu 2008).
We have argued that the standard serializations
of RDF are not well suited for multilingual ontology
editing as such. Special purpose ontology editors
avoid problems by building their own graphical
editing interfaces often borrowing from Eclipse.
Many standalone ontology editors exist, both open
source and commercial.
5 EDITING IN RDF/XHTML
In designing TF, we did not want to build yet
another application. Instead, we wanted to choose or
adapt a serialization format for the web that is
familiar to users, has support in general purpose web
editing tools, without yet compromising machine
processability.
XHTML seems to best fill the bill. As an
extension of XML, it supports Unicode and can be
manipulated with common XML tools. As the native
representation format of browsers, can be depended
on to provide good support for display. XHTML can
be edited with a wide range of standalone tools and
browser extensions. The HTML 5 standard (2012) is
to merge with XHTML and provide built in support
for direct editing.
The RDF/XHTML format represents RDF
models in the form of a sorted HTML list of tree-
structured entries, isomorphic with Turtle. It
supports WYSIWYG editing through a user
definable XHTML skin (Figure 1). The idea is
similar to that applied in XML editors like
XMLmind (2012).
The layout of the XHTML document can be
customised with templates also written in RDF. The
output of the XHTML writer can be varied with a
number of parameters:
KEOD2012-InternationalConferenceonKnowledgeEngineeringandOntologyDevelopment
366
Figure 1: Editing TF/XHTML with a textarea editor.
Table 1: RDF/XHTML parameters.
template=<URI>
RDF template to define the structure of the
serialization
schema=<URI>
RDF/OWL schema to bridge between
TermFactory and user ontologies
root=<URI>
output filter (list of instances/classes to
include in the output)
active=<URI> active ontology (editable triples)
locals=<URI>
localization vocabulary (a multilingual
term ontology)
lang=<ISO
langcode>
localization language
links=<URI>
hyperlink mapping for redirecting resource
URIs
The XHTML writer writes these options it used
in the header of XHTML document head element
and they are stored in the RDF model by the RDF
writer. As a result, an XHTML entry can be
roundtripped through RDF without having to bother
about the settings, and RDF/XHTML entries
complete with layout can be stored in a RDF
database.
TF tries to facilitate multilingual terminology
editing by making terms communicable across
locales. TF does this by way of localization, using
URIs as an interlingua. It converts local terms to
global URIs and localizes them back as terms
familiar to users in another locale (Figure 2).
TF meta classes and properties like Concept,
Term, or hasDesignation are also described in TF as
(instances of) concepts, terms, and properties. This
means that TF is capable of reflection: it can
document and localise itself. A TF schema
localisation ontology provides definitions and
translations of TFS descriptive classes and
properties in TF itself. This information is then used
to change language in the TF front end tools. Figure
2 shows an entry localized into Chinese.
Ontology edit operations, such as actual
modification of the active ontology with the edits are
supported by a webservice API implemented as an
Axis2 webservice EditService using the Jena
library, mediated by a Java servlet EditForm.
6 CONCLUSIONS
RDF/XHTML has no editing support for multiple-
RDF triple constructs like OWL axioms. Currently,
it is best suited for editing RDF or OWL instance
bases (ABox).
HTML 5 defines primitives to support online
editing of HTML documents. We expect to be able
to generalize our approach further come HTML 5.
RDF/XHTML:OntologyEditinginHTML
367
Figure 2: TF/XHTML entry localized into Mandarin.
REFERENCES
Buitelaar, P., Cimiano, P., McCrae, J., Montiel-Ponsoda,
E, Declerck, T. 2011. Ontology Lexicalization: The
lemon perspective, TIA 2011, November 8-10 2011.
http://oa.upm.es/9772/1/Ontology_Lexicalisation.pdf
(accessed 4 July 2012)
Carlson, L., 2012. TermFactory Manual. http://tfs.cc/doc/
TFManual_en.xhtml (accessed 30 April 2012)
Francopoulo G., Bel N., George M., Calzolari M., Pet M.,
Soria C., 2008. Multilingual resources for NLP in the
lexical markup framework (LMF). Language
Resources and Evaluation (revue) ISSN 1574-020X
(print) + 1572-0218 (online) Springer Netherlands
Gutierrez, C., Hurtado, C., Mendelzon, A., Pérez, J., 2011.
Foundations of Semantic Web databases. J. Comput.
Syst. Sci. 77(3): 520-541 (2011)
HTML5, 2012: A vocabulary and associated APIs for
HTML and XHTML. Editor's Draft 28 April
2012.(Accessed 30 April 2012)
Kudashev, I., Carlson, L., Kudasheva, I., 2010.
TermFactory: Collaborative Editing of
TermOntologies. In: Bhreathnach, Ú., Barra-Cusack,
F. (eds.) Terminology and Knowledge Engineering
Conference 2010, pp. 479–500. Fiontar, Dublin City
University, Dublin (2010)
OWL, 2009. OWL 2 Web Ontology Language Document
Overview.W3C Recommendation 27 October 2009.
http://www.w3.org/TR/owl2-overview/ (accessed 30
April 2012)
RDF, 2004. Resource Description Framework (RDF).
http://www.w3.org/RDF/ (accessed 30 April 2012)
RDFS, 2004. RDF Vocabulary Description Language 1.0:
RDF Schema. W3C Recommendation 10 February
2004. http://www.w3.org/TR/rdf-schema/ (accessed
30 April 2012)
Turtle, 2010. Turtle – Terse RDF Triple Language.
http://www.w3.org/TeamSubmission/turtle/ (accessed
30 April 2012)
XMLmind, 2012. XML Editor. http://www.xmlmind.com/
(accessed 4 July 2012)
KEOD2012-InternationalConferenceonKnowledgeEngineeringandOntologyDevelopment
368
