Content uploaded by Mark Alan Musen
Author content
All content in this area was uploaded by Mark Alan Musen on Dec 25, 2013
Content may be subject to copyright.
Semantic Web 11-165 (2011) 1–0 1
IOS Press
WebProtégé:
A Collaborative Ontology Editor and
Knowledge Acquisition Tool for the Web
Editor(s): Harith Alani, Knowledge Media institute, The Open University, UK
Solicited review(s): Gianluca Correndo, University of Southampton, UK; Rinke Hoekstra, Vrije Universiteit Amsterdam, The Netherlands;
Philipp Frischmuth, University of Leipzig, Germany; Valentin Zacharias, FZI Forschungszentrum Informatik, Germany
Tania Tudorache ∗, Csongor Nyulas, Natalya F. Noy and Mark A. Musen
Stanford Center for Biomedical Informatics Research, Stanford University, 1265 Welch Road, Stanford, CA 94305,
USA
E-mail: {tudorache, nyulas, noy, musen}@stanford.edu
Abstract. In this paper, we present WebProtégé—a lightweight ontology editor and knowledge acquisition tool for the Web.
With the wide adoption of Web 2.0 platforms and the gradual adoption of ontologies and Semantic Web technologies in the real
world, we need ontology-development tools that are better suited for the novel ways of interacting, constructing and consuming
knowledge. Users today take Web-based content creation and online collaboration for granted. WebProtégé integrates these
features as part of the ontology development process itself. We tried to lower the entry barrier to ontology development by
providing a tool that is accessible from any Web browser, has extensive support for collaboration, and a highly customizable and
pluggable user interface that can be adapted to any level of user expertise. The declarative user interface enabled us to create
custom knowledge-acquisition forms tailored for domain experts. We built WebProtégé using the existing Protégé infrastructure,
which supports collaboration on the back end side, and the Google Web Toolkit for the front end. The generic and extensible
infrastructure allowed us to easily deploy WebProtégé in production settings for several projects. We present the main features
of WebProtégé and its architecture and describe briefly some of its uses for real-world projects. WebProtégé is free and open
source. An online demo is available at http://webprotege.stanford.edu.
Keywords: Web-based ontology editing, knowledge acquisition, collaboration, Protégé, Semantic Web
1. Introduction
Researchers and application developers in many do-
mains have successfully used ontologies to solve a
wide range of problems, including data integration,
configuration, data analysis, annotation, and search [3,
14]. More and more frequently, many of these ontolo-
gies are products of collaborative development, with
active involvement of domain experts, and not just
knowledge engineers. In order to develop tool support
for collaborative ontology development in this setting,
*Corresponding author. Email: tudorache@stanford.edu
we have analyzed the ontology-development process
in a large number of projects developing biomedical
ontologies [21]. We believe that our findings are com-
mon to other domains, as well. In several projects that
we have analyzed, a small group of ontology experts
develops the core of the ontology. After they identify
the main structure and modeling patterns, domain ex-
perts contribute content to the ontology. There are two
main challenges at the step of involving domain ex-
perts in the ontology development: First, the domain
experts must get familiar with knowledge modeling, or
even formal underpinnings of the representation lan-
guage, and ontology-editing tools. These tools are of-
1570-0844/11/$27.50 c
2011 – IOS Press and the authors. All rights reserved
2T. Tudorache et al. / WebProtégé: A Collaborative Ontology Editor and Knowledge Acquisition Tool for the Web
ten complex [4] and create a high entry barrier that is
not easy for domain experts to surpass. Second, while
many users are familiar with collaboration technolo-
gies in the style of Web 2.0 applications, we must
adapt these approaches to ontology development and
make them an integral part of the development pro-
cess, improving the user experience and productivity.
In this paper, we present WebProtégé—a collabora-
tive Web-based platform that supports ontology edit-
ing and knowledge acquisition, and that can be eas-
ily tailored for domain-expert use. While our previ-
ous papers [18,17,19] have focused on particular as-
pects of WebProtégé in the context of deploying the
tool in a number of real-world use cases, this paper
gives an overview of the WebProtégé generic features,
both functional and architectural, and describes how
the tool can be extended and customized to support
other use cases.
The paper is organized as follows. Section 2 gives an
overview of the existing Web-based ontology editors.
We then describe the main features of WebProtégé, in-
cluding its declarative user interface, the support for
collaboration and ontology reuse, and its extensibility
in Section 3. Section 4 gives a high level overview of
the tool architecture. We describe the real-world de-
ployments of WebProtégé in Section 5, and then dis-
cuss current limitations and future plans in Section 6.
2. Related Tools
There are a small number of collaborative ontology
editors that exist today for the Web. Semantic wikis
[8] add semantic capabilities to the traditional wikis.
Most of the semantic wikis focus on enhancing the
content with semantic links that allows a more mean-
ingful navigation and supports richer queries. The se-
mantic wikis usually associate a page to a particu-
lar instance in the ontology, and the semantic anno-
tations are converted into properties of that instance.
For WebProtégé, our focus is on editing the ontology
structure itself (schema, or class-level, TBox), includ-
ing more complex representations (e.g., editing OWL
restrictions), and on knowledge acquisition where the
immediate validation of the values based on the defini-
tions in the ontology is possible. OntoWiki [1] is one
example of a semantic wiki that supports distributed
knowledge engineering, but it also focuses more on
acquisition of instance data and not the ontology or
schema itself. MoKi1[6] is a collaborative tool for en-
terprise modeling, implemented on top of a Wiki that
has been successfully deployed in a number of real
world use cases. MoKi supports the editing of OWL
domain models using the Wiki forms. Its focus re-
mains, however, on supporting the easy modeling of
business processes and enterprise models, rather than
being a fully-fledged OWL editor. Neologism2[2] is a
web-based vocabulary editor and publishing tool that
focuses on building RDF and lightweight OWL vocab-
ularies. As the authors of the tool stress on their web-
site, Neologism is not an ontology editor. The tool of-
fers primitive collaboration features, as it is still work
in progress. Knoodl3is a commercial ontology edi-
tor, built on top of a wiki platform that provides basic
ontology editing features. Knoodl combines the struc-
tured ontology information with a free-text wiki page
and focuses more on searching capabilities and link-
ing to SPARQL endpoints. Soboleo4[23] and Pool-
Party5[15] are Web-based tools for creating collabora-
tively SKOS and RDF vocabularies. They support the
lightweight editing of taxonomies, and their focus is
on providing services that take advantage of these vo-
cabularies, such as annotation or tagging of resources,
faceted browsing, and semantic search.
None of the Web-based ontology tools that we have
investigated provide such extensive editing support for
both the class level and instance level information as
WebProtégé does. We could also not find Web-based
tools that provide customized views of the ontology for
different users, or extensive collaboration support, or
such an extensible and pluggable architecture. A num-
ber of collaborative ontology editors are available as
desktop applications, but they do not work in a Web
browser and require installation on the user’s machine.
A good review of these tools is available elsewhere
[11].
3. WebProtégé Features
WebProtégé6is a Web-based lightweight ontology
editor. Our goal in building this tool was not to offer
yet another ontology editor, but rather to fill in a signif-
1https://moki.fbk.eu
2http://neologism.deri.ie/
3http://knoodl.com
4http://www.soboleo.com/
5http://poolparty.biz/
6http://tinyurl.com/webprotege
T. Tudorache et al. / WebProtégé: A Collaborative Ontology Editor and Knowledge Acquisition Tool for the Web 3
icant gap in the landscape of ontology tools. Our aim
is to provide an ontology tool that a large spectrum
of users, ranging from ontology experts to domain ex-
perts, can use. Thus, the ability to customize the user
interface for users with different levels of expertise was
of utmost importance in our design decisions (see Sec-
tion 3.1). Furthermore, most projects use community-
based approaches to develop ontologies. Section 3.2
describes the main collaboration support in WebPro-
tégé. Ontology reuse is also common in mature ontol-
ogy domains, such as biomedicine. Section 3.3 gives
an overview of how WebProtégé supports the reuse
and interlinking of biomedical ontologies. Creating an
open, extensible and flexible infrastructure that can be
easily adapted to the needs of different projects was
also a top priority in our design (see Section 3.4).
3.1. User Interface
The WebProtégé user interface (Figure 1) is built as
aportal, inspired by similar infrastructures, such as,
iGoogle7and myYahoo,8which are already familiar to
online users. The WebProtégé portal is composed of
tabs, either pre-defined or defined by the users. A tab is
an empty container made up of smaller components—
portlets—that provide independent pieces of function-
ality. WebProtégé comes with a library of portlets, as
well as predefined tabs that are familiar to users from
desktop ontology editors (see Section 3.1.1). A user
can customize the appearance of WebProtégé by rear-
ranging portlets using drag-n-drop, by adding and re-
moving portlets and tabs from the top toolbar. The user
may also save a particular layout configuration for a
project, thus creating a personalized view of the ontol-
ogy. The customized layout will be restored the next
time the user opens the project in WebProtégé.
3.1.1. The Portlet Library
WebProtégé includes a set of predefined tabs,9
which contain the most popular functionality in the
Protégé desktop editor [12]. For example, the pre-
defined Classes Tab enables users to browse and edit
the class hierarchy and the properties of classes. The
Properties Tab provides access to the details of the
properties in the ontology. The Individuals Tab con-
tains forms for acquiring instances of classes.
Besides the pre-defined tabs, users may add their
own tabs containing portlets that are useful for a par-
7http://www.google.com/ig
8http://my.yahoo.com/
9http://tinyurl.com/webprotege-ug
ticular task. They can also re-configure the pre-defined
tabs, removing the portlets that are not useful in their
projects and adding other ones.
Users create their own tabs and custom-tailor the
pre-defined tabs using a library of WebProtégé portlets,
which support common browsing and editing patterns
in ontology development. WebProtégé currently sup-
ports the full-fledged editing of classes, properties and
individuals. For example, the Class Tree portlet, the
Property Tree portlet and the Individuals List portlet
support the creation, deletion and searching of the re-
spective entities. The Properties Portlet displays and
edits the data, object and annotation properties of on-
tology entities (classes, properties or individuals) and
can be used in combination with other portlets. The
Properties View portlet is very popular and displays
the properties that have the selected class in its do-
main. Users find this portlet very intuitive, and inter-
pret its content as seeing the “relationships” of the
class. The Restrictions portlet allows the editing of
the asserted class conditions and includes support for
auto-complete.
We implemented portlets for supporting common
modeling patterns. For example, the Instance Table
portlet, which displays instance values in the form of a
table (each row is an instance, and columns correspond
to properties of the instance), can be used for browsing
and editing reified relations. Several projects currently
use this portlet.
Other useful portlets include the HTML portlet,
which allows the embedding of arbitrary HTML snip-
pets as part of a tab (for example, we currently use
this portlet to include a Twitter feed frame in one of
the tabs), the Metrics portlet, which provides essen-
tial statistics about the ontology, or the Ontology List
portlet, which lists the available ontologies for the cur-
rently logged in user.
All portlets available in WebProtégé can be config-
ured by means of a property list, which gives great
flexibility in customizing a portlet for a particular task.
For instance, the Class Tree portlet can be configured
to display only a certain branch in an ontology, or the
Instance List portlet to display only the instances of
a particular class. Similarly, the buttons, their labels,
and toolbars are controlled through a property list.
For example, we hide the ontology toolbar for certain
projects to discourage users from changing the user
interface configuration, at the request of the ontology
owners.
4T. Tudorache et al. / WebProtégé: A Collaborative Ontology Editor and Knowledge Acquisition Tool for the Web
Fig. 1. The user interface of WebProtégé showing the Ontology for Parasitic Lifecycle (OPL). The user interface consists of several tabs (pages),
such as Classes, Properties, Change History, etc. Each page contains several portlets (e.g., Class tree, Properties, Restrictions, Notes, etc.). A
user can add new tabs and portlets from the toolbar and save the configuration.
3.1.2. Knowledge Acquisition Forms
WebProtégé is not only an editor for classes and
their properties, but also provides extensive support for
knowledge acquisition of instance data. Three of the
real-world projects that we describe in Section 5 show
how WebProtégé can be customized into a knowledge-
acquisition tool suited for domain experts. The cus-
tomized tool presents the domain experts with sim-
ple forms known from other Web-based applications,
without them being really aware that they are, in fact,
editing instances of an ontology.
Figure 2 shows how we configured the knowledge-
acquisition forms for two projects on which we collab-
orate with the World Health Organization. The forms
are created as part of the Property Form portlet,10
which allows the association of a property in the ontol-
ogy to a form field (e.g., textfields, checkboxes, drop-
down lists, radio buttons, etc.) used for displaying and
acquiring the values of that property.
10http://tinyurl.com/webprotege- forms
Forms, similarly to portlets, are configurable using
property lists. All aspects of a form field (label, associ-
ated property, size, etc.) are configurable, including the
groups of users who are allowed to edit that property
(other users will only be able to browse the values).
This configuration enables us to define fine grained ac-
cess control at the level of a property in the ontology.
3.1.3. User Interface Configuration
One of our main goals in developing WebProtégé
was to have a tool that can be easily configured for
different settings and types of users. All aspects of
the WebProtégé user interface (layout, portlets, forms,
etc.) are configured in a XML file11 stored on the
server side. The layout of tabs and portlets can be done
in the user interface directly; the users can drag and
drop various portlets and then save the configuration.
However, the configuration of forms (Section 3.1.2)
currently requires changes in the XML file.12
11http://tinyurl.com/webprotege- uiconfig
12The configuration of forms through the user interface will be
part of a future release.
T. Tudorache et al. / WebProtégé: A Collaborative Ontology Editor and Knowledge Acquisition Tool for the Web 5
Fig. 2. Knowledge-acquisition forms in WebProtégé tailored for domain experts showing the forms for the International Classification of Diseases
(left) and the International Classification of Traditional Medicine (right). The WebProtégé forms allow the creation of customized user interfaces
for specific projects by simply changing the layout configuration.
The declarative user interface allows us to define
custom views for different users. Currently, WebPro-
tégé supports three levels of user interface configura-
tions:
1. A general, default configuration that applies to
all projects with no customizations;
2. A configuration for a specific project;
3. A configuration for a specific project and a spe-
cific user.
WebProtégé builds the user interface dynamically
when the user opens a project by trying to find the most
customized configuration (Option 3), and, if does not
find it, it will fall back to the previous two options.
This flexible mechanism allowed us to build cus-
tom ontology views for different user types. For ex-
ample, a content expert may see the knowledge acqui-
sition forms for a particular branch in the ontology,
while a project manager will be interested in tracking
the changes and activity statistics at the level of the
entire ontology.
3.2. Collaboration Support
WebProtégé provides extensive collaboration sup-
port, including change tracking, contextualized threaded
discussions, watches and notifications, an extensible
access policy mechanism, and generation of statis-
tics of the ontology-development process [18]. Even
though, we reused the functionalities of Collaborative
Protégé [21], which are themselves implemented using
Semantic Web technologies, we still had to add new
functionalities.
All authoring operations in the WebProtégé are
tracked as instances of a Changes and Annotation
Ontology (ChAO) [9]. Several portlets present this
declarative change-tracking information in a user-
friendly way. For example, the Changes portlet shows
the changes performed on an entity: the author of a
change, timestamp, and a user-friendly description of
a change. A different portlet displays the changes per-
formed in the entire ontology.
Users can also have contextualized threaded discus-
sions and notes attached to different entities in the on-
tology. The notes are typed and structured and are also
stored as instances of ChAO. A user may add the Notes
Tree portlet to any tab, and the portlet will display the
notes attached to the selected entity in a threaded view.
For instance, we are using the Notes Tree portlet both
in the Classes Tab and in the Properties Tab, and it will
display the notes attached to classes or properties in
the respective tabs. Figure 1 shows the visual indica-
tion of notes attached to classes as a commenting icon.
We are also displaying the note counts of subclasses
using a smaller icon, so that it is easier for users to
quickly assess the most active ontology branches and
“drill-down” to the relevant classes.
Awatch functionality allows users to express their
interest in certain entities, and unlike wikis, in branches
in the ontology. The Watched Entities portlet will dis-
play the changes and new notes of the watched enti-
ties and branches. A notification mechanism will send
email notifications to users containing direct Web links
to the watched entities that have changed or have new
notes attached to them (This feature is one of the most
appreciated ones in WebProtégé that keeps users en-
6T. Tudorache et al. / WebProtégé: A Collaborative Ontology Editor and Knowledge Acquisition Tool for the Web
gaged). Users may configure their profile to receive
notifications immediately, hourly, daily, or not at all.
Project managers are able to follow the progress of
the development process by using one of the change-
analysis plugins available in the Web platform and also
as a plugin of the Protégé desktop client.13 These plu-
gins present statistics of edits, which can be filtered by
authors and time frame, as well as author-network de-
pendencies, and tag clouds [5].
WebProtégé supports a flexible access policy mech-
anism. We defined some common access policies that
the user interface enforces, such as Read,Write,Create
new users or Display in ontology list. The latter per-
mission controls if an ontology should show up in the
list of available ontologies displayed on the home page
of WebProtégé, and it allows us to make certain on-
tologies private. Administrators may also define other
custom access policies, such as the right to enable or
disable certain tabs or portlets, or to use certain func-
tionalities in the tool. We use an ontology to store the
configuration of the projects available in WebProtégé
and their access policies [20].14 As mentioned in Sec-
tion 3.1.2, WebProtégé also supports a property-level
access policy that allows only a certain group of users
to edit that property in a knowledge-acquisition form.
3.3. Reuse of Biomedical Ontologies
Reusing terms and interlinking of ontologies are
best practices in ontology development. As biomedicine
is one of the domains in which most public mature
ontologies exist, we provide a generic way of reusing
terms from biomedical ontologies stored in the Bio-
Portal repository [10].15 BioPortal is a repository of
over 250 biomedical terminologies and ontologies that
serves a large community, and that can be accessed
both in a Web browser and through RESTful Web ser-
vices. Even though BioPortal is a biomedical resource,
the technology behind the tool is domain-independent
and has already been deployed in several other do-
mains. By accessing the REST services, the BioPortal
Reference portlet allows WebProtégé users to search
for terms in BioPortal ontologies in the context of their
ontology editing, to browse their details, and then to
create a reference to these terms from their evolving
ontology with a single click. The WebProtégé cus-
tomization used for the authoring of the International
13http://tinyurl.com/change- analysis
14http://tinyurl.com/metaproject
15http://bioportal.bioontology.org
Classification of Diseases makes heavy use of the in-
terlinking feature: In the last year, users have created
over 40,000 links to terms in BioPortal ontologies.
3.4. Extensibility
We implemented WebProtégé as a pluggable and
extensible architecture that can be customized to the
needs of any particular project. The plugin infrastruc-
ture on the front end side enables developers to eas-
ily build their own portlets, forms and tabs by imple-
menting predefined Java interfaces. We also tried to
make the plugin interfaces similar to the Protégé desk-
top client ones, to lower the entry barrier for existing
Protégé plugin developers.
Currently, WebProtégé loads any ontology language
and format that is supported in the Protégé 3.x series
(OWL 1.0, RDF(S), and Frames). Developers can add
support for other ontology formats by implementing an
Ontology Service interface16 that separates the server
and the client (see Section 4). In this way, they can
easily add different back ends (e.g., OWL-API, Jena,
triplestores) without having to make any change to the
user interface code. As a proof of concept, and to ad-
dress a common user request, we have already imple-
mented an OWL-API backend for WebProtégé that is
available in the Protégé SVN repository17 and that we
will release soon (see Section 6). The service layer
adds a level of separation between different compo-
nents of the system. The services18, such as the On-
tology Service, the Notes Service, Project Configu-
ration service, and so on, dictate the functionality of
WebProtégé. Changing the implementation of one ser-
vice, does not imply changing the other services. For
example, if one develops a backend for WebProtégé
by implementing the Ontology Service interface, he
or she will not have to provide a new implementa-
tion of the Notes service. This lego-style architecture
allows the combination of components from different
software libraries. For example, the OWL-API back-
end may work with the Protégé 3 notes and discussions
component.
4. Architecture of WebProtégé
Figure 3 shows a high level overview of the WebPro-
tégé architecture. The user interacts with a client ap-
16http://tinyurl.com/wp- ontologyService
17http://tinyurl.com/wp- svn1
18http://tinyurl.com/wp- services
T. Tudorache et al. / WebProtégé: A Collaborative Ontology Editor and Knowledge Acquisition Tool for the Web 7
Fig. 3. The architecture of WebProtégé. The user interacts with the
client applications, such as WebProtégé in a Web browser, or Collab-
orative Protégé on the desktop. WebProtégé and other Web clients
connect to a servlet engine that contains the application server logic.
All clients will eventually connect to a common collaborative frame-
work that provides services such as change tracking, storing of notes
and discussion, access control, and so on.
plication (front end) that runs in a Web browser and is
implemented in JavaScript. The server side (back end)
runs in a servlet container, such as Tomcat, and is im-
plemented in Java.
The WebProtégé front end and back end interact
via a service layer defined as Java interfaces. We cre-
ated interfaces for accessing and changing the ontol-
ogy content (e.g., get subclasses of a class), for change
tracking (e.g., get all changes of an ontology entity),
for project administration and access control, and for
project layout and configuration (e.g., get the layout
for project NCI for user X). The ontology access ser-
vices are currently implemented using the Protégé 3
API in the released version (an OWL-API implemen-
tation is also available, but not released, yet). As we
mentioned in Section 3.4, other back ends can be eas-
ily plugged in by providing an implementation of the
respective services. The WebProtégé services can be
called also by third-party applications.
The WebProtégé back end connects to a Protégé
server that provides access to the ontology, storage,
collaboration features, change management, and all
other back end functionality. The Protégé collabora-
tive framework [21] supports simultaneous editing of
an ontology by multiple Web-based or desktop clients,
tracks the ontology changes, manages the notes and
discussions, and so on. The framework also enforces
the access policies, such as read and write. One ex-
tremely useful feature in this architecture is the fact
that WebProtégé and the Protégé desktop client may
access the same ontology for concurrent reading and
writing. All changes made by either of the clients are
immediately visible in the other clients. This mecha-
nism is important if certain features are not available
in the Web client, for example, reasoning. One of the
projects we host on the WebProtégé demo site, makes
all ontology edits in WebProtégé and it performs the
reasoning on the same copy of the ontology in a Pro-
tégé desktop client.
The WebProtégé front end is implemented using the
Google Web Toolkit (GWT).19 GWT allows develop-
ers to write Java code for the user interface and then
compiles it into optimized Javascript to run in a Web
browser. One huge advantage of GWT is that it enables
Java developers with no Web-based UI experience to
write UI code directly in Java and to use the extensive
support of the Java development environments. GWT
is free and open source and used by a large community
around the world.
5. Use Cases
We deployed WebProtégé in a number of real-world
projects. The most prominent one is the development
of the 11th revision of the International Classification
of Disease (ICD-11) lead by the World Health Organi-
zation (WHO) [18,17,19]. Most of the United Nations
member countries use ICD to compile health statistics,
to monitor health-related spending and to inform pol-
icy makers. Hence, ICD is an essential health-care re-
source around the world.
We created a customized knowledge-acquisition
form tailored for the WHO domain experts (left side
of Figure 2 shows a part of it) by creating a specific
layout configuration for WebProtégé. This project has
served as a requirements’ driver for the development
of several of the portlets and forms currently avail-
able in WebProtégé, such as the BioPortal Reference
portlet. As the requirements and the core ontology (to
which forms were bound) were evolving in parallel at
a rather accelerated pace, we had to make all portlets
very configurable so that we can adapt the user in-
terface quickly. We also implemented some custom
portlets and forms as extensions to the generic ones
(e.g., the ICD class tree portlet is an extension of the
generic class tree portlet, but uses customized icons for
classes based on a displayStatus object property).
19http://code.google.com/webtoolkit/
8T. Tudorache et al. / WebProtégé: A Collaborative Ontology Editor and Knowledge Acquisition Tool for the Web
We branded the customized WebProtégé for ICD-
11 as iCAT.20 iCAT has been in production use since
October 2009. The platform has 200 registered users
from all over the world. Most of the users are in-
ternational medical experts that are not knowledge-
able about OWL or ontologies. They interact with
the knowledge-acquisition forms to enter information
about diseases and to create the new ICD-11 classifi-
cation. The ICD ontology has currently over 5 million
triples, and contains over 31,000 classes that are ac-
tively edited. Users have already made over 105,000
changes in iCAT, and created more than 40,000 cross-
links to external ontologies by using the BioPortal Ref-
erence portlet. More than 19,000 notes and discussions
are recorded in the platform. The Changes and Anno-
tation Ontology (ChAO) used for storing the change
tracking and notes instances has over 5 million triples.
iCAT is under very active use and it will support the
ICD-11 revision until 2015, when WHO will officially
release ICD-11. After 2015, iCAT will be used to sup-
port minor revisions to ICD-11.
Similarly to iCAT, we deployed two other produc-
tion platforms used for the development of other WHO
classifications: the International Classification of Tra-
ditional Medicine (ICTM) and the International Clas-
sification of Patient Safety (ICPS). The right-hand
side of Figure 2, shows a partial screenshot of the
ICTM platform, in which we had to make sure that
WebProtégé works properly with international content,
as ICTM is concurrently authored in 4 languages: En-
glish, Chinese, Japanese and Korean. Both platforms
are in current active use in a production setting.
The WebProtégé demo platform21 hosts several real-
world projects. These projects are developed by groups
of researchers from different domains and we have
configured the access permissions to these ontolo-
gies based on the preferences of the ontology authors.
Therefore, some of the ontologies can be browsed pub-
licly, while others are available only to a closed group
of users. Some of the publicly accessible ontologies
are: the Ontology of Parasite Lifecycle (OPL) devel-
oped by a group of researchers from University of
Pennsylvania and other institutions, an extensions of
the Ontology for Biomedical Investigations (OBI) for
Web service annotations, the Product-Service Systems
ontology, and the NIH Health Indicators. These on-
tologies use different ontology languages (most use
20A demo platform is available at: http://icatdemo.
stanford.edu.
21Available at: http://webprotege.stanford.edu.
OWL, but some are using Frames), and they differ in
the type of edits they are performing. For example,
OPL developers heavily edit class restrictions, and use
knowledge-acquisition forms for authoring the classes
metadata, while developers of other ontologies follow
a more lightweight approach and edit only the class
hierarchy and the class annotation properties.
As anyone can freely download and install WebPro-
tégé on any machine, we do not have exact counts of
external installations or users of WebProtégé. We use
the mailing lists as a gauge of interest and we do get
regular emails about WebProtégé.
6. Discussion and Future Work
We performed several usability studies of WebPro-
tégé and published the results elsewhere [18,17,19].
These results were all encouraging. We have also cre-
ated a usability questionnaire and published it on the
WebProtégé wiki.22 We have posted a request on the
Protégé mailing list to help the Protégé team evalu-
ate our software. We asked respondents to fill out a
questionnaire on various aspects of the Protégé tool. In
the questionnaire, 13% of respondents (19) indicated
that they use WebProtégé. We have contacted these
19 users and asked them to fill out another question-
naire, this new one specifically for WebProtégé. 18 of
them responded. We have received further responses
to the survey from users who have followed the link
directly from the WebProtégé wiki page. Even though,
we omitted to ask about the amount of experience users
had with Protégé or WebProtégé, we concluded from
the responses we received that most respondents have
used WebProtégé for a real project, rather than just try-
ing out the online demo. In this survey, we were mainly
interested in learning whether users find WebProtégé
easy to use, and which features they appreciate the
most and which features are missing. We also used the
responses of the survey to guide our future develop-
ment plans for WebProtégé. We received in total 28
responses to the online questionnaire. 88% of respon-
dents found the tool easy to use, and 75% of them
found it easy to learn. When we asked what the re-
spondents particularly liked about WebProtégé, they
highlighted the fact that the interface is web-based and
there is no installation required. When we asked what
was missing, many users listed the features that were
22http://tinyurl.com/wp- survey2
T. Tudorache et al. / WebProtégé: A Collaborative Ontology Editor and Knowledge Acquisition Tool for the Web 9
available in the desktop version of Protégé, and said
that they wanted all of those features in the Web ver-
sion.
One missing feature that came up several times is
the support for OWL 2. We already implemented an
OWL-API back end [7] for WebProtégé that covers
most of the current WebProtégé services, and we plan
to release it in the very near future. Using an OWL-
API back end will also open up possibilities that were
difficult to support in the current architecture.
For example, several of the user studies, and our
other interactions with users highlighted the need for
users to download a local copy of the ontology and
to edit it off-line. There are challenges in implement-
ing this “sandbox” feature because we must be able to
fold back the changes into the master version. Thus,
we must address cases such as a user creating a sub-
class of a certain class in the sandbox and someone
else deleting this class from the ontology in the mas-
ter version. We envision developing different strate-
gies for merging and in certain cases asking the users
to resolve the conflicts manually (e.g., similar to the
work of Stojanovic [16]). We have already made sig-
nificant steps in addressing this challenge. Our group
has implemented an OWL-API server that supports
the concurrent access to an ontology similar to source
control repositories, such as SVN, using check-in and
check-out operations [13]. The server has an extensi-
ble conflict-resolution mechanism. We plan on using
the OWL-API server and its conflict-resolution mech-
anism to support ontology snapshots and sandboxes in
WebProtégé.
When we started WebProtégé, we envisioned the
WebProtégé interface as a lightweight version of the
interface in the Protégé desktop client. However, the
usability studies and our interactions with users indi-
cate that there is not a feature in the desktop client
that some user does not want in the Web client. We
have already added features to WebProtégé that can-
not be considered light weight (e.g., restriction editing,
the SWRL tab) because users were requesting it. We
are gradually adding the more complex features to the
Web client as users express a need for them, while at
the same time trying to preserve the default lightweight
feel of the Web tool.
Another area of active research in our group is
studying how we can represent better the declarative
user interface of WebProtégé. In the current implemen-
tation, we are using an XML file to configure the user
interface. This approach is reasonable for projects that
do not require complex configurations. However, the
WebProtégé configuration file for ICD is over 8000
lines long (a class has over 45 fields with multiple
configuration), and this configuration is getting very
hard to manage, maintain, and validate. To address
this issue, we created an OWL representation of the
user interface23 [22]. This ontology allows us to cre-
ate templates of configurations, to associate constraints
for configuration parameters and constraints or depen-
dencies between portlets (e.g., the restriction portlet
should be used only with OWL ontologies). We hope
that this approach will make the management and val-
idation of UI configurations easier.
We are currently working on integrating additional
collaboration mechanisms in the platform and analyz-
ing the dynamics of collaborative ontology develop-
ment. We already performed a quantitative and qual-
itative analysis on the change tracking and notes ac-
tivity on several collaborative projects that are using
WebProtégé, and we identified emerging user roles that
are common to all projects [5]. Users with these differ-
ent roles have different interests and different require-
ments for the user interface. As we learn more about
the way domain experts develop ontologies in a dis-
tributed setting, we can adjust the tools to support col-
laboration even better.
As the Web access gets more wide spread, and the
modern Web browsers providing rapid improvements
of their JavaScript engines, we can envision that some
users might prefer to use only the Web version of Pro-
tégé, rather than the “classic” desktop client. Never-
theless, we believe that both the Web and the desktop
client have their own role, and complement each other,
in a similar way that Web and desktop email clients co-
exist and fulfill different requirements. We will con-
tinue to develop in parallel both WebProtégé and the
Protégé desktop client, and we will strive to make them
interoperate seamlessly.
7. Conclusions
We presented WebProtégé—a lightweight ontology
editor and knowledge-acquisition tool for the Web.
WebProtégé provides extensive collaboration features
that are integrated as part of the ontology develop-
ment itself. The tool also comes with highly cus-
tomizable and declarative user interface that can be
adapted to any level of user expertise, as we have
23The ontology is available for browsing on the WebProtégé demo
server, called the UI Ontology.
10 T. Tudorache et al. / WebProtégé: A Collaborative Ontology Editor and Knowledge Acquisition Tool for the Web
proven in the customizations for several real-world
projects. Our experience in using WebProtégé in the
projects run by our collaborators as well as the feed-
back that we received from our larger user community,
show that WebProtégé is actively used, fills in a niche
that no other ontology-development and knowledge-
acquisition tool does, and its user interface is simple
enough to be used by domain experts.
Acknowledgements
We are grateful to Jennifer Vendetti and Timothy
Redmond for contributing code to the WebProtégé user
interface and its back end. We are also very thank-
ful to the users of WebProtégé and of the Protégé
mailing lists who have provided us with requirements
and feedback for the tool. Our sincere thanks to Alex
Skrenchuk for providing system administration sup-
port for the WebProtégé servers. The work presented
in this paper is partially supported by the NLM Grant
1R01GM086587-01. Protégé is a national resource
supported by grant LM007885 from the U.S. National
Library of Medicine.
References
[1] S. Auer, S. Dietzold, and T. Riechert. OntoWiki–a tool for so-
cial, semantic collaboration. In I. Cruz, S. Decker, D. Alle-
mang, C. Preist, D. Schwabe, P. Mika, M. Uschold, and
L. Aroyo, editors, Fifth International Semantic Web Confer-
ence, ISWC, volume 4273 of Lecture Notes in Computer Sci-
ence, pages 736–749. Springer Berlin / Heidelberg, 2006.
[2] C. Basca, S. Corlosquet, R. Cyganiak, S. Fernández, and
T. Schandl. Neologism: Easy vocabulary publishing. In
C. Bizer, S. Auer, G. A. Grimnes, and T. Heath, editors, The
4th Workshop on Scripting for the Semantic Web, at the Eu-
ropean Semantic Web Conference (ESWC-2008), volume 368,
Tenerife, Spain, 2008. CEUR.
[3] O. Bodenreider and R. Stevens. Bio-ontologies: current trends
and future directions. Briefings in Bioinformatics, 7(3):256–
274, 2006.
[4] M. Dzbor, E. Motta, C. Buil, J. M. Gomez, O. Görlitz, and
H. Lewen. Developing ontologies in OWL: an observational
study. In B. C. Grau, P. Hitzler, C. Shankey, and E. Wallace,
editors, OWL: Experiences and Directions, OWLED-2006, vol-
ume 216, Athens, Georgia, US, 2006. CEUR.
[5] S. Falconer, T. Tudorache, and N. Noy. An analysis of collabo-
rative patterns in large-scale ontology development projects. In
M. A. Musen and O. Corcho, editors, K-CAP ’11: Proceedings
of the sixth international conference on Knowledge capture,
pages 25–32, Banff, Alberta, Canada, 2011. ACM.
[6] C. Ghidini, B. Kump, S. Lindstaedt, N. Mahbub, V. Pammer,
M. Rospocher, and L. Serafini. MoKi: The enterprise mod-
elling wiki. In L. Aroyo, P. Traverso, F. Ciravegna, P. Cimi-
ano, T. Heath, E. Hyvonen, R. Mizoguchi, E. Oren, M. Sabou,
and E. Simperl, editors, Proceedings of the European Se-
mantic Web Conference (ESWC-2009), volume 5554 of Lec-
ture Notes in Computer Science, pages 831–835, Heraklion,
Greece, 2009. Springer Berlin / Heidelberg.
[7] M. Horridge and S. Bechhofer. The OWL API: A Java API for
OWL ontologies. Semantic Web Journal, 2(1):11–21, 2011.
[8] M. Krotzsch, D. Vrandecic, and M. Volkel. Semantic Me-
diaWiki. In I. Cruz, S. Decker, D. Allemang, C. Preist,
D. Schwabe, P. Mika, M. Uschold, and L. Aroyo, editors, The
5th International Semantic Web Conference (ISWC06), volume
4273 of Lecture Notes in Computer Science, pages 935–942,
Athens, GA, USA, 2006. Springer Berlin / Heidelberg.
[9] N. Noy, A. Chugh, W. Liu, and M. Musen. A framework for
ontology evolution in collaborative environments. In I. Cruz,
S. Decker, D. Allemang, C. Preist, D. Schwabe, P. Mika,
M. Uschold, and L. Aroyo, editors, The 5th International Se-
mantic Web Conference (ISWC-06), volume 4273 of Lecture
Notes in Computer Science, pages 544–558, Athens, GA, US,
2006. Springer Berlin / Heidelberg.
[10] N. Noy, N. Shah, P. Whetzel, B. Dai, M. Dorf, N. Griffith,
C. Jonquet, D. Rubin, M. Storey, C. Chute, et al. BioPortal: on-
tologies and integrated data resources at the click of a mouse.
Nucleic Acids Research, 37(suppl 2):W170, 2009.
[11] N. F. Noy, H. Alani, G. Stumme, P. Mika, Y. Sure, and D. Vran-
decic, editors. Workshop on Social and Collaborative Con-
struction of Structured Knowledge (CKC 2007) at WWW 2007,
Banff, Canada. CEUR-WS.
[12] N. F. Noy, M. Sintek, S. Decker, M. Crubézy, R. W. Ferger-
son, and M. A. Musen. Creating Semantic Web contents with
Protégé-2000. IEEE Intelligent Systems special issue on Se-
mantic Web Teconology, 16(2):60–71, 2001.
[13] T. Redmond, M. Smith, N. Drummond, and T. Tudorache.
Managing change: An ontology version control system. In
K. Clark and P. F. Patel-Schneider, editors, OWL: Experiences
and Directions, 5th International Workshop, OWLED 2008,
volume 496, Karlsruhe, Germany, 2008. CEUR.
[14] D. Rubin, N. Shah, and N. Noy. Biomedical ontologies: a
functional perspective. Briefings in Bioinformatics, 9(1):75–
90, 2008.
[15] T. Schandl and A. Blumauer. PoolParty: SKOS thesaurus
management utilizing linked data. In L. Aroyo, G. Antoniou,
E. Hyvonen, A. ten Teije, H. Stuckenschmidt, L. Cabral, and
T. Tudorache, editors, The Extended Semantic Web Conference
(ESWC-2010), volume 6089 of Lecture Notes in Computer
Science, pages 421–425, Heraklion, Greece, 2010. Springer
Berlin / Heidelberg.
[16] L. Stojanovic. Methods and Tools for Ontology Evolution. PhD
thesis, University of Karlsruhe, Germany, 2004.
[17] T. Tudorache, S. Falconer, N. F. Noy, C. Nyulas, B. T. Üstün,
M.-A. Storey, and M. A. Musen. Ontology Development for
the Masses: Creating ICD-11 in WebProtégé. In P. Cimiano
and H. Pinto, editors, 17th International Conference on Knowl-
edge Engineering (EKAW-2010): Knowledge Engineering and
Management by the Masses, volume 6317 of Lecture Notes
in Computer Science, pages 74–89, Lisbon, Portugal, 2010.
Springer Berlin / Heidelberg.
[18] T. Tudorache, S. Falconer, C. Nyulas, N. Noy, and M. Musen.
Will Semantic Web Technologies Work for the Development
of ICD-11? In P. Patel-Schneider, Y. Pan, P. Hitzler, P. Mika,
L. Zhang, J. Pan, I. Horrocks, and B. Glimm, editors, The 9th
T. Tudorache et al. / WebProtégé: A Collaborative Ontology Editor and Knowledge Acquisition Tool for the Web 11
International Semantic Web Conference (ISWC 2010), volume
6497 of Lecture Notes in Computer Science, pages 257–272,
Shanghai, China, 2010. Springer Berlin / Heidelberg.
[19] T. Tudorache, S. Falconer, C. Nyulas, M. Storey, T. Üstün, and
M. Musen. Supporting the collaborative authoring of ICD-11
with WebProtégé. In American Medical Informatics Associ-
ation (AMIA) Annual Symposium Proceedings, volume 2010,
pages 802–806. American Medical Informatics Association,
2010.
[20] T. Tudorache and M. Musen. Collaborative development of
large-scale biomedical ontologies. In S. Ekins, M. A. Z.
Hupcey, and A. J. Williams, editors, Collaborative Computa-
tional Technologies for Biomedical Research, Technologies for
the Pharmaceutical Industry, pages 179–200. John Wiley and
Sons, Hoboken, New Jersey, 2011.
[21] T. Tudorache, N. Noy, S. Tu, and M. Musen. Supporting Col-
laborative Ontology Development in Protégé. In A. Sheth,
S. Staab, M. Dean, M. Paolucci, D. Maynard, T. Finin, and
K. Thirunarayan, editors, Seventh International Semantic Web
Conference, ISWC 2008, volume 5318 of Lecture Notes in
Computer Science, pages 17–32, Karlsruhe, Germany, 2008.
Springer Berlin / Heidelberg.
[22] T. Tudorache, N. F. Noy, S. M. Falconer, and M. A. Musen.
A knowledge base driven user interface for collaborative on-
tology development. In P. Pu, M. J. Pazzani, E. Andre, and
D. Riecken, editors, Proceedings of the 16th International
Conference on Intelligent User Interfaces, IUI ’11, pages 411–
414, Palo Alto, CA, USA, 2011. ACM.
[23] V. Zacharias and S. Braun. SOBOLEO-social bookmark-
ing and lighweight engineering of ontologies. In N. F. Noy,
H. Alani, G. Stumme, P. Mika, Y. Sure, and D. Vrandecic, ed-
itors, Workshop on Social and Collaborative Construction of
Structured Knowledge at the 16th International World Wide
Web Conference (CKC 2007), volume 273, Banff, Canada,
2007. CEUR.
