An Editorial Workflow Approach For Collaborative
Ontology Development
Rau´l Palma1, Peter Haase2, Oscar Corcho1, Asuncio´n Go´mez-Pe´rez1, and Qiu Ji2
1Ontology Engineering Group, Laboratorio de Inteligencia Artificial
Facultad de Informa´tica, Universidad Polite´cnica de Madrid, Spain
{rpalma,ocorcho,asun}@fi.upm.es
2Institute AIFB, University of Karlsruhe, Germany
{pha,qiji}@aifb.uni-karlsruhe.de
Abstract. The widespread use of ontologies in the last years has raised new
challenges for their development and maintenance. Ontology development has
transformed from a process normally performed by one ontology engineer into
a process performed collaboratively by a team of ontology engineers, who may
be geographically distributed and play different roles. For example, editors may
propose changes, while authoritative users approve or reject them following a
well defined process. This process, however, has only been partially addressed by
existing ontology development methods, methodologies, and tool support. Fur-
thermore, in a distributed environment where ontology editors may be working
on local copies of the same ontology, strategies should be in place to ensure that
changes in one copy are reflected in all of them. In this paper, we propose a
workflow-based model for the collaborative development of ontologies in dis-
tributed environments and describe the components required to support them. We
illustrate our model with a test case in the fishery domain from the United Nations
Food and Agriculture Organisation (FAO).
1 Introduction
The growing use and application of ontologies in the last years has lead to an increased
interest of researchers in the development of ontologies, either from scratch or by
reusing existing ones. Ontology development and maintenance activities are addressed
by many different methodologies (e.g. Methontology, On-To-Knowledge, DILIGENT,
etc.). However, most of them only consider the development of ontologies by single
users or a small group of ontology engineers placed in the same location. More impor-
tant is that even though they address the methodological aspects, in general they focus
less on the process followed by organisations to coordinate the collaborative ontology
development. In practice ontologies may be distributed, and a whole team of ontology
engineers with different roles may collaborate in the development and maintenance,
usually following a well defined process. Examples of such collaborative development
processes can be found in international institutions like the United Nations Food and
Agriculture Organisation (FAO), who are developing and maintaining large ontologies
in the fishery domain [8]. Other similar examples are those of the Gene Ontology (GO)
J. Domingue and C. Anutariya (Eds.): ASWC 2008, LNCS 5367, pp. 227–241, 2008.
c
© Springer-Verlag Berlin Heidelberg 2008

228 R. Palma et al.
project1, which addresses the need for consistent descriptions of gene products in dif-
ferent databases, the caGrid project2, which aims at providing a virtual informatics
infrastructure that connects data, research tools, scientists, and organizations, etc.
Consequently, in this collaborative organisational setting, existing approaches are
not enough to support all ontology development and maintenance needs. Furthermore,
although recently some proposals and tools have been designed specifically to sup-
port collaborative ontology development (e.g. client-server mode in Prote´ge´ along with
the PROMPT and change-management plugins), they generally only address parts of
the overall problem (see section 4). Most of the existing advanced ontology tools (e.g.
Prote´ge´ core system, SWOOP, etc.) support only the single-user scenario, where there
is just one user involved in the development and later modification of the ontologies.
With such tools, a typical scenario of collaborative ontology development would look as
follows: An editor changes an ontology using his ontology editor system and then sends
(e.g. using email or uploading it to an ontology repository) his locally changed ontol-
ogy to other users (i.e. to add more changes using their own Prote´ge´ system, or review
current changes). Even in the scenario where all users are editing the same ontology
stored in a central server (e.g. using client-server mode in Prote´ge´), the coordination of
the actions of the editors (e.g. when editors want their changes to be reviewed or what
kind of actions they can perform) is not yet fully supported.
As we can see from the previous discussion, in this type of collaborative scenario,
change management is central. Hence, we need appropriate procedures (and corre-
sponding infrastructure) to control and support the management of ontology changes.
This procedure can be modelled as a collaborative workflow, which according to [2],
is a special case of epistemic workflow characterized by the ultimate goal of designing
networked ontologies and by specific relations among designers, ontology elements,
and collaborative tasks. The need for such workflows has also been acknowledged in
the past by other related works (e.g. [17]). An example of such workflow is that fol-
lowed by the FAO (described in [8]), which we take as a use case in our work, in order
to derive a generic set of required activities to support it.
Following this workflow, the development process starts with proposals for ontology
changes. These proposals are discussed by multiple users (with different roles) in a col-
laborative way. For instance, if a change is made by an ontology editor, it has to be ap-
proved by a validator. After that, the change will be considered definitive and permanently
added to the structure. Once changes are definitive, we will have a new stable version of
the ontology, which requires the appropriate support to manage different ontology ver-
sions. Of course one could think of other kinds of workflows in different situations.
In this paper we present our approach for the management of collaborative ontol-
ogy development in a distributed scenario by means of an editorial workflow. We
analyse the collaborative development process using as illustrating scenario the case
study at FAO and derive a set of functional requirements to support the process. We
then introduce our proposal to support the collaborative ontology development where
we address the identified problems. In particular, we propose a formal model for the
representation of the workflow and describe the relationship with other models and
1 http://www.geneontology.org/
2 http://www.cagrid.org/