Publishing OWL ontologies with Presto
Alexander DeLeon1 and Michel Dumontier1,2
1 School of Computer Science
2 Department of Biology
Carleton University, 1125 Colonel By Drive, Ottawa, Ontario, K1S5B6 Canada
adlbatti@scs.carleton.ca, michel dumontier@carleton.ca
Abstract. Publishing RDF/OWL ontologies on the Semantic Web typ-
ically starts by placing the document in a web accessible location and
ends with redirects of ontological components (classes, properties, indi-
viduals) to that the document. Unfortunately, this is seldom sufficient for
expressive OWL ontologies in which reasoning is essential to determine
the full extent of the entity in question. Moreover, the ability to dy-
namically query expressive ontologies yields new applications over static
publishing including the possibility that these queries may be equivalent
to new ontological entities. Here, we describe the design of a new tool for
publishing OWL ontologies in a dynamic manner such that the ontology
and all of its entities are web resolvable and queryable, hence opening
new avenues for knowledge management on the Semantic Web.
1 Introduction
A core objective of the Semantic Web is to add machine understandable de-
scriptions to the current Web, in part, by publishing ontologies that describe
and relate entities using formal, logic-based representations. An essential aspect
of the Semantic Web is to ensure that the terminology defined in ontologies are
web-accessible such that information about the ontological entity may be dis-
covered and links with related entities explored. The Linked Data architecture
[1] suggests that HTTP URIs may be used as resource names, whether they be
electronic documents or conceptual representations (i.e. the class of Cat or the
first author). Should HTTP URIs be web resolvable then web client may discover
additional knowledge by following links between these web-accessible resources.
The W3C Semantic Web Deployment Working Group recently proposed best
practices in the publishing of RDF vocabulary on the web [2] in which the resolu-
tion of named entities (classes, properties, individuals) is achieved by redirection
to a single document in which axiomatic statements are made about it. This ap-
proach may be exceedingly cumbersome should the entity be defined in a very
large document, which is generally the case in ontologies for the life sciences.
Crucially, the proposed solution may not be appropriate for OWL ontologies as
the knowledge about an entity depends on the cumulative sum of knowledge
obtained from the import of OWL documents into the knowledge base. Thus,
the description of some entity may differ from one knowledge base to another.

Dynamic generation of ontology documentation may be provided using the
OWLDoc server3 whose interactive user interface with reasoner-enabled query
capabilities reveal the sophistication of OWL ontologies. While OWLDoc server
generates project-specific HTML documentation with permanent links for shar-
ing, it does not currently provide RDF/OWL descriptions for use on the Seman-
tic Web.
In this paper, we describe Presto (Published RESTful Ontology), a tool for
publishing and querying OWL ontologies on the Semantic Web. Presto allows any
party to publish their OWL knowledge base, including any imported documents.
For a given ontology, Presto provides the following:
1. A self-referential namespace for all ontological documents and entities, so as
to follow linked knowledge as a static ontological snapshot.
2. A RESTful service for DL and SPARQL queries that are identified by per-
manent HTTP URIs.
3. Content-negotiation capabilities to retrieve dynamically generated HTML or
RDF/XML.
Finally, we demonstrate Presto’s value in i) maintaining ontological interoper-
ability, ii) building new ontologies with terminology defined from queries of other
ontologies and iii) ontology version control.
2 Overview
Figure 1 illustrates the general architecture of Presto, a Java based applica-
tion whose central data model relies on the OWL API (version 2.1.1) [3]. The
general operation is as follows. Publishers invoke Presto with an OWL docu-
ment and a target publishable URI that should resolve to the machine where
Presto is running. The Presto Manager is a mediator component that provides
RESTful services by interacting with several system components including a
DL-Reasoner, a lucene-based entity indexer [4], Manchester Syntax parser from
Protege4 and HTML rendering from OWLDoc.5 The Restlet Framework6 is used
to create HTTP handlers and representations for the URIs of the ontology and
its entities. For instance, if the request Accept header includes the mime-type
application/rdf+xml the response is rendered using RDF/XML otherwise an
HMTL document generated with OWLDoc is returned to the client. The HTTP
handler of the ontology URI has a special behavior which is that it could act
as a query endpoint. When a request is made to the ontology URI using the
query parameter, the server executes the query on the ontology, and returns the
results as OWL, XML or HTML depending on the Accept header and the query
language used. This querying service is futher described in section 4.
3 http://www.co-ode.org/downloads/owldoc-server/
4 http://protege.stanford.edu
5 http://www.co-ode.org/downloads/owldoc/
6 http://www.restlet.org/