AN AJAX APPLICATION TO IMPROVE ONLINE

ONTOLOGY BROWSING

Ontology Explorer

Matteo Busanelli, Massimo Bondi and Nicola Gessa

ENEA Bologna, Via Martiri di Monte Sole 4, Bologna, Italy

Keywords: OWL, Ontology, AJAX, GWT, Browser, Semantic Web, Web Usability.

Abstract: OWL ontologies are one of the fundamental bricks of the Semantic Web architecture. In spite of this,

ontology visualization on a web browser is nowadays not so easy; commercial browsers just show

ontologies in plain text, with the OWL code. To avoid this, many ontology developers have implemented,

together with their own ontologies, specific web applications to display them. But this can not be the right

direction: a common Internet user does not have an easy way to browse an online ontology. Starting from a

wide analysis of these applications for ontology visualization, considering their limitations and the last

solutions for the development of dynamic web tools, the authors propose a GWT based architecture that,

exploiting AJAX technologies, can ease the interfacing of web users with OWL ontologies.

1 INTRODUCTION

Starting from the first paper that delineated the

vision of the Semantic Web (Berners-Lee et al.,

2001) many efforts have been done to produce

specifications, technologies, guidelines and so on to

develop this vision, where data could be understood

by automatic applications. The architecture of the

Semantic Web nowadays is strongly based on the

definition of a set of technologies, (XML, RDF,

etc.). Among these, OWL, a language for ontology

definition, is surely one of the fundamental.

Ontologies have been defined in different ways.

One of them is “An ontology is an explicit

specification of a conceptualization” (Gruber, 1995);

ontologies should then represent the “knowledge”

below plain data. They can perhaps be considered as

the meeting point between the human vision and the

automatic applications. For this reason new tools

should be provided to help users that are non expert

in software or informative tools, to approach the

Semantic Web technologies.

Starting from these considerations, our work

targets to the development of a web-based tool that

could represent the right support for those that do

not know (and that also do not want to know)

anything about the technologies underling

ontologies, but that can obtain from them serious

advantages in their activities. After the analysis of

the limits of the main common software for ontology

management, we present an innovative web

application, named Ontology Explorer, for the

ontology browsing that, avoiding installation or

configuration activities, aims to improve the

usability and comprehensibility of ontologies that

are published online. In our perspective, ontology

browsing is a relevant feature that should be well

supported by a software architecture that aims to

improve the ontology and Semantic Web usability.

In the next section we will present the state of art

about web applications for ontology browsing.

Section 3 will present our design approach for the

development of the application. In section 4 we will

more deeply describe the implemented interface and

the functionalities, detailing the innovative aspects

we have introduced in respect of other solutions.

In chapter 5 we will describe our first

conclusions and the future developments.

2 WEB APPLICATIONS FOR

ONTOLOGY BROWSING

The first consideration starting this work was that

ontologies are written in OWL file, and this format

is not well supported by browsers like Internet

Explorer™ or Opera™, that are not able to represent

OWL files in a convenient manner, but instead in

Busanelli M., Bondi M. and Gessa N. (2009).

AN AJAX APPLICATION TO IMPROVE ONLINE ONTOLOGY BROWSING - Ontology Explorer.

In Proceedings of the International Conference on Knowledge Management and Information Sharing, pages 83-88

DOI: 10.5220/0002270200830088

 SciTePress

general show the plain XML representation.

Studying the state of art, our main references

have been a set of web applications for ontology

browsing and Protégé, which is a widespread java-

based desktop application for ontology management.

Regarding the web applications, we have

selected and analysed a set of 9 products. They are:

• Owl to Ace (Kaljurand, 2006): transforms

an OWL ontology in Attempto Controlled

English (ACE). It can be considered, instead of a

browser, like an application that translates an

ontology from a machine-readable form in a

human-readable one.

• HarmonISA Ontology Viewer (Hall,

2006

), Csml Ontology Viewer (Jeong et al.,

2006), Nature Navigator (the guide - What is

Nature Navigator?, 2007), IASCF Taxonomy

Viewer (International Accounting Standards

Committee Foundation., 2005): allow the

visualisation and browsing of a predefined

ontology. It is based on a set of html pages

linked each other.

• RxNav (Lussier, & Bodenreider, 2007):

allows the browsing of a specific predefined

ontology about pharmaceutical products. It is

based on a java application that is invoked by the

link on a web page.

• Ontology Lookup Service (Cote et al.,

2006): allows the visualisation and browsing of a

predefined ontology about genetics. It displays

the information in a tree structure, representing a

taxonomy that is the starting point to access to

cards about the elements of the ontology.

• GoFish (Berriz et al., 2003): allows the

visualisation and browsing of a set of specific

predefined ontologies about genetics. It is

implemented in an application that displays the

information in structures like trees, cards or

tables.

• Pellet (Sirin et al., 2003): it acts more like

an ontology analyser than an ontology browser,

managing any kind of ontology (RDF, OWL). It

runs in batch mode, and there is no runtime

interaction with the user. It allows to specify the

target ontology. The analysis consists in a plain

text file that reports the extracted information.

• KSMSA Ontology Browser (Ševčenko,

2003): it is a web application for browsing the

SUMO (Niles & Pease, 2001) ontology and all

the connections with WorldNet, developed

within the KSMSA project. In particular this tool

is an online version of the more powerful editor

that could be downloaded. All the ontologies are

defined in the KIF (Knowledge Interchange

Format) language. The browser is strictly

developed for this set of ontologies.

Clearly, this list reports only the main aspects of

each tool. The main considerations about these

applications, starting from which we will fix a set of

requirements for the Ontology Explorer, are:

• These tools do not allow the browsing of an

ontology specified by the user.

• These tools are not configurable, in any

perspective. This means that the user cannot

customise the interface and some functions in

order to make the tool more useful.

• Sometimes, the description level is too

high, with too many unclear technical details,

making it too cumbersome and discouraging.

• In many of them there is not the possibility

to execute research operations on the ontology.

It is worth to note that the wide amount of web

tools to allow Internet users to visualize online

ontologies demonstrates the need for the ontology

designers to show on the web in a useful manner

their work. In other words, up to now, many

ontology developers not only design their own

ontology, but also they implement brand new

“specific” tools to allow the consultation of their

ontologies. In many cases this approach tries just to

solve isolated issues, without considering the online

visualisation of ontologies as an issue that is

common for ontology designers, and without

pursuing a unique solution.

After this first analysis, we concentrated on the

characteristics of Protégé, that is not a web

application, but that perhaps represents the main

reference in the ontology definition field.

We have identified the following drawbacks in

ontology browsing using Protégé (we refer to

version 3.2):

• The use of labels is inadequate, and quite

accessory. The browsing functionalities are

based on the use of IDs, which are used to

provide semantic information to the users.

• The interface for class management is

separated from the interface for instance

management. This presupposes in the user the

knowledge of the differences between classes

and instances;we want to avoid such assumption.

• Class visualisation: Protégé imposes the

distinction between the logical and the property

view; this is not intuitive for a common user, and

again requires specific skills. Moreover, the

default visualisation is the logical one, that does

not display the list of the properties for a selected

class, that we consider one of the fundamental

information that is sought about classes.

KMIS 2009 - International Conference on Knowledge Management and Information Sharing

• The research functionality acts separately

on classes and instances: it is impossible to do a

research on both classes and instances from the

same entry point. Moreover, the research is

performed only on IDs, and not on labels. Again,

this is a strong limitation, and prevents the

definition of multi-language ontology (since

research can be performed only in one language).

The research interface for instances is also

laborious and not easy to use.

• Instance visualisation is not well-structured.

• The descriptions, both for classes and for

instances, are displayed in tables with

property/value keys, and are reported with the

label rdfs:comment. Such descriptions could be

instead provided in a more clear way (for

example using tooltips on the labels).

Some of these limitations will be faced in the next

releases of Protégé. Finally, we considered also

SWOOP and Longwell, but these are not web-

applications, and they are out of our context.

3 THE ONTOLOGY EXPLORER

Differently by other well-recognized widespread

formats in Internet, OWL/XML files are always

displayed by web browsers as plain XML files.

Although we did not develop a plug-in, we wanted a

way to visualize in the browsers (not in the OWL

code) OWL/XML ontologies that could be found in

Internet, without the need to download specific

software, install it, etc. etc.

The Ontology Explorer is designed to be

intuitive to use (also for the inexpert user) and many

visualization and navigation configuration

alternatives are available. During the design phase,

we first considered the limitations listed in section 2.

Then we have outlined for the tool some use cases

that have been identified in different projects (Moda-

ML (Gessa et al., 2005), DDTA Puglia (Regione

Puglia DDTA, 2006) and Leapfrog IP (LEAPFROG-

IP project)) for the development of solutions for

business interoperability and we identified the

generic requirements and features for the software.

3.1 Use Cases for Ontology Browsing

Without going to much in details, one of the results

of the Moda-ML project has been the definition of a

vocabulary of terms upon which a set of XML

document schemas has been defined for data

exchange in the Textile/Clothing sector (De Sabbata

et al., 2005). For this vocabulary we built a semantic

representation (defined with a set of OWL ontology)

of both the terms and the documents defined in the

vocabulary (Gessa et al., 2006).

On the other hand, the DDTA project

(Digitalizzazione dei Distretti a supporto della

filiera produttiva del Tessile/Abbigliamento)

supports the definition and implementation of

services to enable the constitution of enterprise

networks within the textile supply chain in Puglia.

This project exploits the results of Moda-ML. In

these contexts domain experts are fundamental.

Usually, domain experts have great knowledge about

concepts that concern their expertise area, but their

knowledge about ontology implementation is quite

absent. Then, an example of a typical user of the

Ontology Explorer (OE) could be a textile expert

who consults the Moda-ML vocabulary and

documents from a semantic point of view.

In this scenario of B2B interoperability:

1. It has been identified the need to search for

specific concepts defined in a domain ontology

in order to understand their properties and to

design the proper applications and data formats

to manage and exchange the related information.

2. In order to manage XML Schema

documents, it has been identified the need for the

designer to search for semantic concepts

modelled in the XML Schema documents. In this

case we need a tool to easily browse among the

concepts treated in the documents, without

studying the document structure. This procedure

is useful to identify reusable components already

defined in existing documents.

3.2 Requirements for the Tool

We have identified a set of non-functional

requirements for the tool to develop. The tool

• must be able to load, read and visualize every

ontology that is written in OWL.

• must hide the technical and formal aspects

related to the ontology language.

• should be as much as possible interactive,

allowing a fast and friendly use.

• should be configurable at different levels.

• must provide a powerful and easy to use search

mechanism, regardless the way the concepts

have been formalised (for example, if they are

classes or instances).

• must be able to visualize classes and instances

in parallel and in a homogeneous way.

The Ontology Explorer allows the loading of an

OWL ontology, to browse and display every kind of

AN AJAX APPLICATION TO IMPROVE ONLINE ONTOLOGY BROWSING - Ontology Explorer

information related to the concepts defined in the

ontology and to perform a search on the items of the

ontology. In the following sections the capabilities

of the application will be detailed deeply.

3.3 Architecture of the Application

In order to achieve the objectives fixed during the

analysis phase, we adopted the AJAX technology,

which allows developing highly interactive web-

applications, exploiting asynchronous

communication between client and server. The

architecture of the tool is composed of:

• A service (server side) that implements the

OWL logic and manage the loaded ontology; we

called it Ontology Service.

• A set of beans-like Java objects, called

OWL Wrappers, that have only to carry the

information extracted from the ontology between

the service and client.

• A client interface running on the web

browser that is based on GWT AJAX dynamic

widgets plus CSS style sheets for layout

formatting.

The client will represent all the information

coded in the wrappers using graphical components

that we have called Semantic Widgets. These

widgets are derived from the basic ones that GWT

offers (like tree, textbox, etc…). They have been

built to implement and exploit the logic of many

OWL elements (i.e. taxonomy or relations between

instances). The wrappers are used to populate the

widgets with the needed content. In some cases, a

different content “forces” the same widget to assume

a specific behaviour: the behaviour of a widget

depends on the characteristics of the content to

display. The main idea behind these Semantic

Widgets is to have a library of reusable graphical

components that could “understand ontological

structures” and to use them to better represent

information and to improve the user experience.

All the Semantic Widgets were developed in a

way that makes them easily graphically

customizable, allowing the setting of colours, layout,

images and so on simply using CSS stylesheets.

4 INTERFACE DESIGN

The interface of the Ontology Explorer

(http://www.cross-lab.it/cross-lab/imple/pgcl.asp?

lingua=en&p=247&node_id=6.1.1) has been

structured in three different sections, in order to well

separate and identify the different operations that

can be performed, to reduce the confusion and the

embarrassment for the user and to improve the

understandability of the information.

4.1 The Loading Section

In the top of the program interface an area is

dedicated to the management of the application.

Here is the form to load the target ontology, to

provide general information about it, to configure

some aspects related to the browsing and finally to

show error messages if need.

There are two ways to start the Ontology

Explorer and to load a remote ontology in the tool:

• In the first way by connecting to the tool

(http://winter.bologna.enea.it/OntologyExplorer/

OntologyExplorer.html) and then specifying the

URL of the OWL file in the text box.

•

The second way allows to visualize in the

Ontology Explorer, linked directly by another

web source (like an HTML page), a specific

concepts within its ontology. In other words,

appending to the URL of the application,

respectively the URL of the OWL file of an

ontology and the name of the concept you are

looking for (i.e. adding to http://

winter.bologna.enea.it/OntologyExplorer/Ontolo

gyExplorer.html the string “?ontologyUrl=

ontologyURL&concept =conceptName”), the

Ontology Explorer will be open directly on the

conceptName.

Figure 1: The interface of the Ontology Explorer.

When an ontology is already loaded (as in Fig. 1;

due to space constraints, we show only a small

image), here the user can find the ‘More Info’ button

that opens a pop-up panel showing all the general

information about the loaded ontology. Also, the

user can find the ‘New Ontology’ button that is useful

to load a new ontology.

KMIS 2009 - International Conference on Knowledge Management and Information Sharing

4.2 The Navigation Section

The left part of the interface is dedicated to show a

dynamic taxonomy tree used to browse and manage

the taxonomies defined by the ontology. Once the

ontology has been loaded, the tree is filled with all

the information related to the taxonomies. The

taxonomies can show both the classes and the

instances or only classes. This aspect is in any case

configurable (see about application configuration in

the following). In this part of the interface it is also

available a “search field” that allows to search in the

taxonomy tree the desired concepts. The search is

dynamic in the sense that while the user is typing

text in the search box, if he presses the “Enter” key,

a dynamic list of all the concepts matching the typed

text is presented. In this way the user could change

the search on the fly and, in many cases, he doesn’t

have to complete the text that he is looking for. In

fact, by clicking on the list and pressing the ‘Go’

button, the tree will be opened at the concept level in

the taxonomy and the corresponding detail panel

will be displayed in the description section. Since

classes and instances are both visualised in the tree,

in order to distinguish them we represent them with

different formatting styles: in particular the classes

are always presented in bold, while instances are

presented in italic. Another important feature of the

tree is the tip that, if enabled by the configuration,

shows a brief textual description (if exists) of the

concept every time the mouse pointer passes over it.

4.3 The Description Section

The right part of the interface is dedicated to the

visualisation, in specific tab panels, of the details

related to the concepts (classes or instances) selected

in the taxonomy tree. In fact, by clicking an element

in the tree, a tab is opened here. If the user clicks on

a class the opened tab panel will show the class

name (label), the description, the super class and two

sub-panels: one for the list of the properties (each

one containing its description and other property

information like domain and range) and another for

the list of the instances (each one containing its

description tip and being clickable to open the

relative detail panel). If the user clicks on an

instance the opened panel will show, together with

the name and the description, also the list of the

property values for the instance. This part of the

browser can collect at the same time many tabs to

hold information about different selected objects.

The informative elements in the panel are all

clickable, in order to open directly from here other

tabs and to allow a more powerful navigation

between the different concepts crossing the whole

ontology. Together with this interface, other kinds of

information are provided using pop-up, in order to

not fill the interface with too many data.

4.4 Configuration

In our perspective, an ontology can be used by

different kinds of persons, with different skills. We

consider very important to allow the configuration of

our tool in order to match with the browsing

requirements of the different users.

We then predisposed our application to provide a

set of configuration parameters. For this purpose

there is a panel where the user can decide if to show

the classes along with their instances or not, and to

show labels instead of names (OWL IDs) for a better

explanation and research of the concepts modelled in

the ontology. The configuration section of the

Ontology Explorer consists in a pop-up that appears

when the user clicks on the ‘Options’ button.

Through this pop-up the user can configure many

visualization aspects of the OE:

• To show or not instances of classes in the tree

(the search will be performed also among

instances accordingly to this choice).

• To show or not tooltips with the description (if

any) of the concepts.

• To show concepts in the tree by their unique

ontology names (ID) or by their labels.

• To set the search in “start with” (fast search)

mode or in “substring” mode (slow).

• To show instances in a tab of a class by default

using names or, if any, labels.

5 CONCLUSIONS AND FUTURE

WORKS

Nowadays proper tools for online OWL ontology

reading and browsing are lacking, or are limited.

As shown in the use cases we identified in

section 3.1, Semantic Web technologies can solve a

set of problems related to business vocabulary

management. We developed an AJAX-based web

application named Ontology Explorer that

overcomes some of the limitation of existing tools

and allows an intuitive ontology browsing for every

kind of Internet user. In particular, the OE:

1. is able to load every OWL/XML ontology

on the Internet; in this sense it is not built to

manage a specific ontology. We have tested it

AN AJAX APPLICATION TO IMPROVE ONLINE ONTOLOGY BROWSING - Ontology Explorer

with many online ontologies, and with the Moda-

ML ontology, that includes 373 classes, 44

relationships and 1098 instances.

2. matches with the skills of a person that is

not an ontology expert. To this aims, the

concepts of classes, instances, datatype or object

property are completely transparent to the users.

A central issue has been also the design of

friendly forms for information visualisation.

3. provides a powerful research interface.

4. exploits the use of labels, that allows the

browsing of multi-language ontologies, a more

strong comprehension of the terms and a stronger

search and browsing. The concept IDs in an

ontology should be used as are primary keys in a

database, just to identify a class/instance/-

property, but IDs should not be used to provide

semantic information (as usually happens).

It is worth to note that the Ontology Explorer

distinguishes itself from other tools like Protégé

because it should be considered not like an editor for

Ontology Expert (like Protégé), but like a powerful

browser for Domain Expert. Considering our use

cases, exploiting the semantic representation of the

Moda-ML business vocabulary, the tool has

simplified the interfacing with the terms and the

XML Schema documents, allowing a more easy

search and analysis of them. The future steps to

improve the Ontology Explorer will concern mainly:

• To enhance the configurability of the tool

providing a new set of customizations about the

domain of user’s interest. This will be achieved

by defining some Semantic Profiles that could be

adopted by the user (externally to the ontology)

to browse a specific ontology.

• To add a new Advanced Semantic Search

that could use logic inference to deduce non

explicit information and that allows enlarging the

resulting set of the matching concepts to all the

semantically related ones.

• To improve the graphical aspect and the

usability and to solve some compatibility issues

with some browsers (for instance Opera™ and

Firefox™ don’t render some widgets so well).

REFERENCES

Berners-Lee, T., Hendler, J., Lassila, O.. The semantic

web. Scientific America, May 17, 2001.

Gruber, T. R.. Toward principles for the design of

ontologies used for knowledge sharing. In

International Journal of Human Computer Studies 43

(5/6), pag. 907–928, 1995.

Kaljurand, K., Fuchs, N.E.. Bidirectional mapping

between OWL DL and Attempto Controlled English.

Budva, Montenegro, 2006.

Hall, M.. HarmonISA Land-Use Viewer System - System

Documentation and Handbook. HarmonISA

Publication, July, 2006.

Jeong, E., Nagasaki, M., Saito, A., Miyano, S., Ontology-

Based Representation for Cell System Markup

Language. CSML Publications, 2006.

Lussier, Y.A.., Bodenreider, O.. Semantic Web:

Rivolutionizing knowledge discovery in the life

sciences. Springer, 2007.

About the guide - What is Nature Navigator?. The Natural

History Museum, 2007.

Cote, R.G., Jones, P., Apweiler, R., Hermjakob, H.. The

ontology lookup service, a lightweight cross-platform

tool for controlled vocabulary queries. BMC

Bioinformatics, 28 February, 2006.

International Accounting Standards Committee

Foundation. About the IFRS-GP Taxonomy Viewer.

International Financial Reporting Standards, General

Purpose Financial Reporting for Profit-Oriented

Entities, Incorporating Additional Requirements for

Banks and Similar Financial Institutions (IFRSGP),

May 15, 2005.

Berriz, G.F., White, J.V., King, O.D., Roth, F.P.. GoFish

finds genes with combinations of Gene Ontology

attributes. Bioinformatics, 2003.

Sirin, E., Parsia, B., Grau, B.C., Kalyanpur, A., Katz, Y..

Pellet: A Pratical OWL-DL Reasoner. Journal of Web

Semantics, 2003.

De Sabbata, P., Gessa, N., Cucchiara, G. , Imolesi, T.,

Vitali, F.. Supporting eBusiness with a dictionary

designed in a vertical standardisation perspective.

IEEE-CEC Conference, Munich, Germany, July 19-22

2005, pp. 160-167, ISBN 0-7695-2277-7.

Gessa, N., Cucchiara, G., De Sabbata, P., Brutti, A.. A

bottom-up approach to build a B2B sectorial standard:

the case of Moda-ML/TexSpin. In Proceedings of

INTEROP ESA 05-NOREST Workshop, Geneve

February 22nd 2005, publication by Herm`es Science

Publications.

Regione Puglia. Assessorato allo Sviluppo Economico,

Progetto ”Distretto Digitale a supporto della filiera

produttiva del Tessile Abbigliamento” (DDTA), 2006.

Gessa, N., Busanelli, M., De Sabbata, P., Vitali, F..

Extracting a semantic view from an ebusiness

vocabulary. IEEE International Conference on E-

Commerce Technology, San Francisco, California,

June 26-29, 2006, pp 398-401, ISBN 0-7695-2511-3.

Ševčenko, M.. Online Presentation of an Upper Ontology.

In Proceedings of Znalosti 2003, Ostrava, Czech Rep.,

February 19-21, 2003, ISBN 80-248-0229-5.

Niles, I. Pease, A.. Towards a Standard Upper Ontology.

2nd International Conference on Formal Ontology in

Information Systems (FOIS), Chris Welty and Barry

Smith, eds, Ogunquit, Maine, October 17-19, 2001.

LEAPFROG-IP project: http://www.leapfrog-eu.org/

LeapfrogIP/main.asp

KMIS 2009 - International Conference on Knowledge Management and Information Sharing