AN ONLINE ONTOLOGY NAVIGATOR FOR WEB

APPLICATIONS IN EDUCATION

Shireesha Tankashala, Fangfang Liu, Brian Horton and Shengru Tu

Computer Science Department, University of New Orleans, New Orleans, LA 70148, U.S.A.

Keywords: Ontology, Semantic Web, Computing Ontology, Information Retrieval, Training, METOC.

Abstract: We have developed a tree-view browser to support development of information retrieval applications that

utilize ontologies. The purpose of this browser is to shield the complexity of an ontology from the users. We

have also reported three ontology-based Web applications in education and training. The scope of these

applications is to cover the daily activities of instructors and trainers. For example, a professor can create

highly customized study guides using a bioinformatics ontology, a trainer can define the training needs for a

specific task based on a network security ontology, and a domain expert can generate the Web forms for

specialized training based on a METOC ontology. These applications were developed using our tree-view

browser. The reuse of the browser component has significantly reduced efforts in the development.

1 INTRODUCTION

Recognizing that ontology is a core component of

the semantic Web architecture (Berners-Lee, 2001),

researchers and practitioners have invested

significant efforts in establishment of ontologies in

many areas. In early years, a methodology for

ontology-based knowledge management was

suggested in (Sure, 2002) which also reported an

early case study of the methodology at a large

enterprise. Happel envisioned a long list of adoption

and utilization of ontologies in all the phases of

software life cycle (Happel, 2006). A large number

of exemplar ontologies in bioinformatics have

demonstrated maturity and completeness (EBI,

2009). As to computer science education, Cassel has

led the ambitious computing ontology project for

many years (Cassel, 2007).

As Davies said: “the ontology itself is only as

valuable as the applications that are developed to

apply it to specific problems.” (Davies, 2006)

Observing the steady growth of the computing

ontology at the “Computing Ontology Project” site

(

http://what.csc.villanova.edu/twiki/bin/view/Main/

OntologyProject), we have realized that it is time to

put more efforts in developing applications using

ontologies in education. In this paper, we shall

report three ontology-based applications in

education and training scenarios. The scope of these

applications covers the daily activities of instructors

and trainers. For example, a professor can create

highly customized study guides using a high-quality

ontology, a trainer can define the initial training

needs for a specific task based on the ontology of the

field, and a domain expert can generate the Web

forms for specialized training. In doing so, we have

developed a tree-view browser that supports the

“navigate, select, produce” three-step process which

can effectively support information retrieval in

various applications. As reported in (Ribaudo,

2009), “download of information” is the dominant

activities in their university’s virtual environment.

We believe information retrieval will be the

common activities in ontology applications.

The contribution of this paper is the effective

approach that shields the complexity of ontology

from the users and ease the development of ontology

applications for education. The primarily targeted

users are the instructors and trainers in the fields

where some reasonable ontologies have been

established. Advanced students and trainees can also

join the user base. The diversified examples in this

paper are to illustrate how effectively information

can be retrieved with the help of ontologies. The

focus of this paper is on the technical support to

educational needs rather than plotting a new

pedagogical technique.

The rest of this paper is organized as the

following. In Section 2 related work is briefly

surveyed. In Section 3, three applications are

319

Tankashala S., Liu F., Horton B. and Tu S. (2010).

AN ONLINE ONTOLOGY NAVIGATOR FOR WEB APPLICATIONS IN EDUCATION.

In Proceedings of the 2nd International Conference on Computer Supported Education, pages 319-324

DOI: 10.5220/0002802003190324

 SciTePress

demonstrated followed by the description of our

special tree-view browser for navigation in

ontologies. The technical details of implementation

are revealed in Section 4. Section 5 concludes.

2 RELATED WORK

The computing ontology project (Cassel, 2007;

Davies, 2006) will have no doubt to take the central

role in the curriculum development in computer

science. A major emphasis of the project is to make

the topics and the relationships between the topics

available to instructors and curriculum developers.

With such a long-term goal, defining relationships is

very prudent. In (Cassel, 2007), only the basic,

simple relationships, such as HasPart, Uses,

Instance, and IsA (including their inverse) are

considered. The has-prerequisite relationship was

suggested in (Khan, 2007). Visualization was

prudently planned for activities such as visualization

of the union of the bodies of knowledge of the

computing disciplines in (Cassel, 2007). Since our

emphasis is on applications of ontologies rather than

establishment of ontologies, we are more flexible in

adding (overlapping) relationships to ontologies in

the way described in Section 4.3.

Bower emphasized an important kind of

taxonomy in computing education: the ten task types

from “declarative tasks” and “comprehension tasks”

to “solve-a-problem tasks” and “self-reflect tasks”

(Bower, 2008). We believe that this learner-oriented

taxonomy can be a supplement to Cassel’s teacher-

oriented ontology.

3 DEVELOPMENT OF

ONTOLOGY-BASED

APPLICATIONS

We have developed a number of online applications

that retrieve information for different purposes from

a set of ontologies. These applications share a

common graphical user interface (GUI) that

facilitates navigation in the ontology and selections

of the items in the ontology. This GUI is a tree-view

browser.

3.1 The Tree-view Browser

A difference between our tree-view browser and

many ontology tools currently available on the

market is that we display not only the subclasses in

the hierarchy, but also the relationships that connect

to other classes in the same tree view. For instance,

in Fig. 2 in addition to show class Asset’s subclasses

“Human” and “Countermeasure” as the “S” nodes,

the relationship “assetHasVulnerability” is also

displayed as an “R” node. Under class

“Countermeasure”, nine relationships such as

“employedAt” and “isContainedIn” lead to seven

targeting classes such as “AssetLifeCyclePhase” and

“Product” respectively in Fig. 1. In the tree view, the

“S” nodes denote the subclasses; the “R” nodes

denote the range classes (the classes connected

through a relationship). The check box associated

with each node allows the user to select the class.

Clicking on a class name, its attributes will be

displayed in a panel (not shown in Figure 1).

Figure 1: A tree view of a fragment of the security

ontology.

Having the class hierarchy displayed, the user

performs a concept-based search (Vieira, 2009).

Having the relationships displayed along with the

“hierarchy”, the user performs a concept-based and

workflow-based search at the same time. This is

much more informative to the viewer who is not

familiar to the given ontology, compared to Protégé

that provides relationships view (property view) in a

view separated from the class hierarchy.

CSEDU 2010 - 2nd International Conference on Computer Supported Education

320

While the ontology is presented in a tree view,

the data structure may contain loops since we have

included the property classes. A range class pointed

by a property (relationship) is possibly a class that

already appeared in the earlier part of the tree. Thus,

the algorithm controlling the tree view display marks

recurrence of classes in order to avoid forming any

infinite loop.

The "Generate" button shown at the bottom of

Figure 2 is the controller that can trigger an action

according to the applications. A typical action is to

generate an XML file that records the selection of

classes by the user, which will be the metadata of the

follow-up processing.

Figure 2: Instant display of a class’s definition.

3.2 Biology Study Guide Producer

This is an application for biology professors to make

customized study guides for students in learning

gene-related subjects. In the bioinformatics fields,

numerous high-quality ontologies have been

established (EBI, 2009). For example, every class in

the gene ontology (

http://www.biopax.org/release

/biopax-level3.owl) has a very informative comment

section which is organized in multiple pieces such as

“Definition”, “Comment” and “Examples”.

The biology professor navigates in the gene on-

tology as shown in Figure 2, and selects the concepts

that need to be included in the study guide. In this

application, we enabled the event handler of the

class nodes. Upon clicking on each class node, the

browser will display the definition, synonyms and

examples in the comment of this class. Thus the

professor can have a preview of the choice. By

checking the checkbox associated to the classes, the

professor can select any number of concepts.

Finally, by clicking the “Generate” button, an e-

book of the selected terms and related information is

generated in the PDF format.

3.3 Network Security Training

Materials Assemble

(a): A selected technology in the security ontology.

(b): Another selected technology in the security ontology.

Figure 3: Selected technologies for their training needs.

AN ONLINE ONTOLOGY NAVIGATOR FOR WEB APPLICATIONS IN EDUCATION

321

Computer network security has been a highly

demanded area of competencies. This is a highly

specialized and highly complex area. Depending on

the context (the hardware, operating systems,

networks, communication protocols) of the system

that holds the task, the required competencies can be

very diverse. In such a case, specifying the training

needs of the given task can require a comprehensive

and deep understanding of the knowledge and skills

in the security area, which can be a significant

challenge to training managers. Using the security

ontology shown in Figure 1, we have produced a

Web application that helps the training managers to

specify the training needs.

For example, a specific task deals with

“X.509CertificateData” and “tripwire”. The training

manager first finds these two targeted technical

topics in the network security ontology by

navigating in the tree view or by searching the

strings. As shown in the two fragments of the tree

view in Figure 3, the two topics,

“X.509CertificateData” and “Tripwire”, are selected.

In this application, the “Generate” button (not

shown in Figure 3) will produce a list of topics

consisting of the name of every ancestor node of the

two chosen topics (“X.509CertificateData” and

“Tripwire”), except for those nodes in the top three

levels. Therefore, the training needs are made of two

groups: (a) “X.509CertificateData”,

“DigitalSignatureData”, “MessageDigestData”,

“CertificateData” and “ElectronicToken”; (b)

“Tripwire”, “Host-BasedIntrusionDetectionSystem”,

“ByBehaviourOnDetection”, “Monitoring”, and

“IntrusionDetectionSystem”.

3.4 METOC Training GUIs Producer

This example came from a project for the training

needs of the Meteorology and Oceanography

(METOC) community (Navy, 2009). The weather

and ocean data are utilized by vastly diversified user

applications that consume a wide range of METOC

data including gridded forecast model data,

climatology, weather effects data, raw satellite data,

space environment and solar, remote-sensed

observations, as well as imagery (METOC, 2009).

A basic and critical training need is to assure the

data entry personnel are able to provide the

applications with correct data. Ideally, a set of GUIs

would be able to mimic the data entry scenarios of

the in-field tasks of data entry. However, there are

simply too many varieties of data request scenarios.

Furthermore, developing highly customized

METOC GUIs is very costly because it is typically

difficult for the GUI programmers to understand the

highly specialized METOC terminologies and

distinguish synonyms and antonyms in different

contexts for many tasks. We developed an ontology-

based application that automates the generation of

highly-tailored training GUI components. The

ontology that supports this application includes a

business-level ontology (BO) which is mostly

consistent with the typical workflows.

Figure 4: The tree-view of the METOC ontology.

To make the training GUI for a data entry task,

the METOC engineer (the domain expert) navigates

in the BO facilitated by our tree-view browser and

selects the nodes of the needed data items as shown

in Figure 4. For this complex application, an

additional feature was added to the data selection

process. Upon the expert’s clicking of a class node,

all its attributes are displayed in a panel on the right-

hand side pane. The expert can deselect any attribute

that is not needed; by default every attribute is

selected. In Figure 4, two classes are selected:

COAMPS and Bounding_Box. In this selection

process, the domain expert can effectively specify

the requirement of the training GUI since the

navigation is in the language of his/her domain

knowledge. As a result, the identifiers of the selected

ontology classes, as well as the selected attributes of

CSEDU 2010 - 2nd International Conference on Computer Supported Education

322

these classes, are recorded in an XML file which

comprises the metadata of the training GUI Web

application.

Upon invoking the training GUI application, the

attributes of the selected nodes are displayed in a

tabular format. Once the user fills in and submits the

form, the data is saved as “individuals” (instances of

the class) in the OWL file by the Web application on

the server side. The GUI for classes COAMPS and

Bounding_Box are shown as Figure 5. As illustrated,

not only the attributes of the chosen classes are

displayed and collected, but the attributes of every

superclass are also included.

4 IMPLEMENTATION

4.1 Implementation of the Tree View

Browser

We have used the Jenkov JSP Tree Tag library

(http://java-source.net/open-source/jsp-tag-

libraries/prize-tags) to iterate over the classes and

build the tree. The tag library consists of a tree

model API. The tree tags display the tree model in

the JSP page. Firstly, a tree instance is created.

ITree tree = new Tree();

The tree instance is to hold the information about

what nodes are expanded or selected. This

information is not kept in the tree nodes. Secondly, a

root tree node is created and it is passed to a method

called drilldown by which the child tree nodes are

added recursively. Finally, the tree instance is given

the root of the tree to be displayed.

onto1 = new OntAccess(filePath, true);

ITreeNode root = new TreeNode(c.getLocalName(),

c.getLocalName(), "root");

The user may drill down the entire tree by

expanding the nodes over the Web. When he/she

clicks on a particular node, the related information is

displayed in a panel on the right-hand side pane. The

information to be displayed can vary depending on

the application. For instance, definitions and

examples are displayed in the case of Section 3.2

where the nodes represent different terms and

concepts in the field of genetics.

The tree-view browser is the user interface that

connects to the ontology model residing on the

server side. This ontology model has the following

six methods which are implemented using Jena API

(

http://jena.sourceforge.net/ DB/index.html).

a. List all the subclasses of each object:

String[] getSubClasses (OntClass C)

b. Get Object Property of each object:

OntProperty[] getObjectProperties(OntClass C)

c. Get range classes of a property:

OntClass[] getRangeClasses(OntProperty p)

d. Return Attributes of a class:

ArrayList getAttributes(OntClass C )

e. Get all of the superclasses:

OntClass[] getSuperClasses(OntClass C)

f. Get individuals of a class:

ArrayList getIndividual(OntClass C)

Figure 5: The GUI for METOC training.

4.2 A Note about Overlapping Local

Additions to a Public Ontology

While modifying a commonly accepted public,

global ontology should go through a rigorous

approval procedure in order to preserve its authority,

local additions to an ontology should be allowed

instantly. For instance, having accessed the

Computing Ontology, a computer science

department wants to specify the prerequisite

relationships among many topics according to the

curriculum of this department. One way to add the

AN ONLINE ONTOLOGY NAVIGATOR FOR WEB APPLICATIONS IN EDUCATION

323

prerequisite relationships to the local copy of the

given ontology is to create a property called

“requires” in the OWL file of the ontology directly,

and then add each prerequisite relationship between

two topics as a sub-property of property “requires”

with proper domain and range classes.

A problem will happen when the computing

ontology is updated. How can we keep the local

copy up-to-date without manual modifications? A

solution is to start with an empty local OWL file,

which then imports the remote computing ontology

into this local ontology. Then the addition of those

properties that represent the prerequisite

relationships will be preserved in the local OWL

file. When the remote ontology is updated, the next

loading of the imported ontology will be the latest

one.

5 CONCLUSIONS

We expect to see more applications showing the

utilizations of ontologies by all kind of users and for

diversified purposes. It is certain that numerous

utilizations will in turn stimulate a fast growth of the

ontologies themselves. Compared to any other

industry, education would most benefit from the

applications of ontologies because knowledge

management and information retrieval are the two

central activities in education.

The reported three applications are in the

category of information retrieval. For these kinds of

applications, our tree-view browser has

demonstrated to be suitable. The tree-view browser

can comfortably load the current Computing

Ontology. The user will not be overwhelmed by the

large number of nodes (classes, or concepts) because

initially most of the nodes are folded. The user has

to open the choice of nodes and then the tree view

shows the result of drill down. Thus as the user

navigates in the ontology, she/he is also controlling

the visibility scope. Level by level, a user knows

where the current focus is. So far our tree-view

browser is capable of handling all the popular

ontologies.

The ontology can be so large that the computer

memory runs out, typically due to many direct and

indirect imports of other ontologies. For that case,

we have used Jena API’s database persistence

capability. That is, the ontology model is held in a

database rather than all residing in the memory. The

form of view will also be changed. We have

developed a multiple-column browser similar to

MSpace (

http://mspace.fm/).

ACKNOWLEDGEMENTS

This work is partially supported by the United States

National Science Foundation grant CCF-0939108.

We sincerely thank Dr. Roy Ladner at Naval

METOC Command, U.S. Navy, for his technical

guidance in the METOC project.

REFERENCES

Berners-Lee, T, Hendler, J., Lassila, O., 2001. The

Semantic Web, Scientific American, the May issue.

Bower, M., 2008. A Taxonomy of Task Types in

Computing, ITiCSE’08, 281-285.

Cassel, L., et al, 2007. The Computing Ontology –

Application in Education, Work group reports on

ITiCSE, 171-183.

Davies, G., Cassel, L.N., Topi, H., 2006. Using a

Computing Ontology for Educational Purposes,

Proceedings of ITiCSE, Bologana, Italy, 334.

EBI, 2009 retrieved. European Bioinformatics Institute

(EBI), Ontology Lookup Service,

http://www.ebi.ac.uk/ontology-lookup/ontologyList.do

Happel, H-J, Seedort, S., 2006. Applications of Ontologies

in Software Engineering, Proceedings of the 2

international workshop on semantic Web enabled

software engineering (SWESE), Athens, GA.

Khan, J, Hardas, M., 2007. Observing Knowledge

Clustering for Educational Resources Using a Course

Ontology, K-CAP’07, Whistler, B.C., Canada, 193-

194.

METOC, 2009. Joint METOC Public Data

Administration, http://www.cffc.navy.mil/metoc/

Navy, 2009 retrieved, Navy Weather Forcasting,

http://hcs-metoc.apl.washington.edu/

forecasting_challenges.html

Ribaudo, M., Rui, 2009. M., AULAWEB, Web-Based

Learning as a Comodity, CSEDU’09, 41-46.

Karlsruhe, Germany, ACM Press, article No. 10.

Sure, Y., Staab, S., Studer, R., 2002. Methodology for

Development and Employment of Ontology based

Knowledge Management Applications, SIGMOD

Record, 31:4, 18-22.

Vieira, A.C. de M., Cruz, S.M.S, 2009. Semantic

Annotations and Retrieval of Pharmacobotanical Data,

CSEDU’09, 333-338.

CSEDU 2010 - 2nd International Conference on Computer Supported Education

324