AN ONTOLOGICAL APPROACH FOR THE ACTIVITY

IN AN AUTOMATIC CONTROL FACULTY

Liliana Dobrica, Alexandra Suzana Cernian and Anca Bertesteanu

Faculty of Automatic Control and Computer Science, “Politehnica” University of Bucharest

313 Splaiul Independentei, Bucharest, Romania

Keywords: Ontology, Natural language processing, OWL.

Abstract: The current work aims at offering a new, ontology-based approach for organizing and retrieving the

information about the educational offer and the activity of the department of Automatic Control of the

Faculty of Automatic Control and Computer Science at the University Politehnica of Bucharest. The project

is divided into 2 main phases. The first phase consists of creating an ontology which contains information

about the disciplines, the teachers, the teaching resources, the research and publication activity in our

faculty, while the second phase consists of developing an application for exploring and querying this

ontology in a controlled natural language.

1 INTRODUCTION

The current work proposes an ontological approach

for modelling the activity in the department of

Automatic Control of the Faculty of Automatic

Control and Computer Science at the University

Politehnica of Bucharest. The main phases of our

project are: creating an ontology which contains

information about the educational, research,

publication and administrative activity in our faculty

and developing an application which explores and

queries this ontology in a controlled natural

language.

The rest of the paper is structured as follows:

Section 2 presents an overview on ontologies;

Section 3 presents our ontological approach,

focusing on the architecture of our system, the

creation of the ontology, the application for

exploring the ontology in natural language and an

evaluation of the application; Section 4 draws the

conclusions of our work and discusses some possible

future developments.

2 ONTOLOGIES BACKGROUND

The origin of the term “ontology” is found in

philosophy, where the result of the efforts to

organize and model the universe is called ontology.

Over time, the term has been borrowed in other

domains as well. In computer science, an ontology is

used to model the knowledge in a specific domain.

It contains the concepts related to that domain, as

well as the relations between them.

The most well known definition of ontologies in

computer science was given by Gruber, in 1993:

“An ontology is an explicit specification of a

conceptualization.” (Gruber, 1993).

An ontology is made up of several main concepts

(Horridge, 2009a), which will be described shortly

in what follows. The classes (also called types,

species or categories) define the concepts of the

domain. A class is a blueprint of an object. The

instances are members of the classes. They can be

actual things, such as persons, objects, animals, etc.,

or even abstract notions. The ontology produces the

taxonomy of its instances. The attributes are

properties of the objects in the ontology. They can

be represented as simple data types (strings,

numbers, boolean or enumerations) or they can be a

reference to another object of the ontology.

Relations indicate how the concepts and the

instances are interconnected and how they

communicate. The entire set of relations in an

ontology defines the semantics of the domain

(Horridge, 2009a).

423

Dobrica L., Suzana Cernian A. and Bertesteanu A..

AN ONTOLOGICAL APPROACH FOR THE ACTIVITY IN AN AUTOMATIC CONTROL FACULTY.

DOI: 10.5220/0003611104230426

In Proceedings of the 6th International Conference on Software and Database Technologies (ICSOFT-2011), pages 423-426

ISBN: 978-989-8425-77-5

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

3 OUR APPROACH

3.1 The Goal

When we started this project, we had in mind a

number of objectives. First of all, we wanted to

create an ontology which models the educational

activity of the department of Automatic Control of

the Faculty of Automatic Control and Computer

Science at the University Politehnica of Bucharest.

The ontology contains information about the

disciplines, the specializations, the teachers, the

teaching resources, the research activities, the

publication activities in our faculty, as well as all the

connections between these elements. After creating

the ontology, we wanted to provide the users the

possibility to visually explore it in a friendly way.

Thus, we implemented an exploration application,

which loads the ontology and allows the users to

browse through the instances and concepts of the

ontology. The last step of the project is to establish a

grammar, based on which the ontology can be

queried in controlled natural language. Both the

ontology and the grammar are easily extensible.

The application is designed in an open and user

friendly manner, so that it can be used by individuals

with limited background knowledge about

ontologies or the OWL language (Antoniou, 2008).

Moreover, the application can load any ontology and

query it in natural language, based on the created

grammar.

3.2 The System Design

Figure 1 presents an overview of the architecture of

the system we propose. Mainly, the ontology

exploration application is a mediator between the

user and the ontology.

Figure 1: Overview of the architecture of the system.

The application accesses the ontology through an

OWL/XML parser from the OWL API (Horridge,

2009b).

Figure 2 illustrates the UML use case diagram

(UML Standard, 2011) of the application.

Figure 2: Use case diagram.

The ontology was created under Protégé 4.1

platform (Horridge, 2009a), with full support for

OWL 2.0 (Antoniou, 2008). Two inference

mechanisms, Hermit (Motik, 2009) and Pellet

(Pellet, 2010), have been used to check the

consistency of the ontology. Both of them can be

installed as Protégé plug-ins.

The application for exploring the ontology was

implemented in Java (Sierra, 2008) and uses several

OWL API and Pellet libraries.

3.3 Creating the Ontology

The ontology models a knowledge base of the

didactic, research and administrative activities in the

department of Automatic Control of the Faculty of

Automatic Control and Computer Science at the

University Politehnica of Bucharest. It covers

information about: disciplines, specializations,

human resources, publications, research activities

and teaching resources. During the creation phase of

the ontology, the following sources have been

consulted: the official web site of the Faculty of

Automatic Control and Computer Science (http://

acs.pub.ro), the official web site of the department of

Automatic Control and Industrial Informatics (http://

www.aii.pub.ro/), the official web site of the

department of Automatic Control and System

Engineering (http://ics.pub.ro/), and the students’

web site (https://studenti.pub.ro).

The main classes of the ontology are presented in

Figure 3.

ICSOFT 2011 - 6th International Conference on Software and Data Technologies

424

Figure 3: The main classes of the ontology.

These classes define the taxonomy and the

vocabulary of the ontology, which will be

intensively used by the queries in natural language.

The classes are disjoint (Horridge, 2009a).

Each of the classes in the ontology has a set of

subclasses and a number of properties. The

connections and the restrictions between these

classes are established through object and date

properties. Figure 4 depicts the set of object property

assertions and data property assertions defined for

the Databases course. The object property assertions

specify when the course is taught, who teaches it, in

what room, what formative category it belongs to, to

which package of courses it belongs and what is the

laboratory room for this course. The data property

assertions establish the criteria for passing the

practical and theoretical exams, the credit point

associated to this course and the number of points

assigned to each activity.

Figure 4: The properties for the Databases course.

Each class has one or more members, which

represent the instances of the ontology. Currently,

the ontology has 67 classes and 253 instances.

3.4 Exploring the Ontology

The exploration application aims at providing a

friendly experience for the users who want to access

the ontology. The application has 3 modules: the

ontology exploration module, the natural language

querying module and the disciplines’ correlations

module. The users do not have the permission to

modify the ontology. Mainly, the exploration part of

the application allows the users to browse through

the classes and the instances of the ontology. When

selecting an instance of the ontology, the users will

visualize in a friendly way all the available details,

as well as the relations the instance has with other

instances. For instance, if a user selects the name of

a teacher, the application will display: the personal

details such as email address, office, telephone

number, the courses taught by that teacher, the

teaching resources linked to him/her, the research

projects he/she is involved in, the books, magazine

or proceedings articles published (with correspond-

ding details).

In what follows, we will discuss the possibility

of querying the ontology in controlled natural

language. The application will interpret the

questions, parse the ontology and provide the

answers to the users. Our approach for natural

language querying is based on RDF triples

(Antoniou, 2008) and it is designed to interpret the

Romanian language syntax, with or without the

punctuation specific to interrogative sentences.

When a user launches a question in natural language,

a string processor class maps it to an RDF triple and

then identifies the subject and the predicate of the

triple. For example, if a user introduces the question:

“What email does Alexandra Cernian have?” (in

Romanian), the subject – Alexandra Cernian and the

predicate - has email, will be identified. Afterward,

the ontology is checked to see if these elements are

found. If they are found, that means that there is a

corresponding triple in the ontology. Finding the

triple leads to finding the missing element (the

object). For our example, it will be found that there

is an instance called AlexandraCernian, which has

the property HasEmail with the value

alexandra.cernian@aii.pub.ro. The questions must

respect a certain grammar, which will be described

in the next paragraph.

The application identifies and interprets the

following types of questions (Cernian, 2011):

AN ONTOLOGICAL APPROACH FOR THE ACTIVITY IN AN AUTOMATIC CONTROL FACULTY

425

• Category membership. The predicate must

contain the “are” or “belong” keywords.

• Figures about instances. The question must

begin with “how many” and the predicate must

be the “has” keyword. It must not contain the

following keywords: “are” and “belong”.

• Relations between instances. The question

must begin with “what” or “who”. The

predicate must contain “has” and must not

contain “are” or “belong”.

• Unknown type. Any question that does not fall

under any of the 3 categories described above

is considered to be of unknown type. In this

case, the system will try to establish if there is

any matching entity in the ontology.

A matching index is used to establish the

correspondance between a term in a key position

within the question and an entity or a relation in the

ontology.

The application has been tested with a set of 75

questions in natural language. The questions belong

to all 4 possible categories descibed above. The

results of the tests proved that the information is

correctly retrived from the ontology. So, the

application is able to correclty interpret natural

language and extract the most relevant information

from the ontology.

4 CONCLUSIONS

This paper presented an ongoing project for an

ontological modelling of the educational, research

and publication activity in the department of

Automatic Control of the Faculty of Automatic

Control and Computer Science at the University

Politehnica of Bucharest. The ontology created

contains information about the human resources,

disciplines, teaching resources, administrative

details about the courses and the personnel, research

activities and publication activities. An application

was also developed for exploring the ontology.

Through this application, the user can browse

through the content of the ontology in a friendly way

and, moreover, he can query the ontology in natural

language, provided he formulates the questions

according to a specific grammar. The ontology is

easily extensible, so it can constantly reflect the

updates in our faculty. The experimental results

were very satisfactory and encourage us to pursue

with improving the informational content of the

ontology and of the application. As a future plan, we

also intend to extend the grammar, in order to

improve the natural language processing

capabilities.

ACKNOWLEDGEMENTS

This work was supported by CNCSIS – UEFISCSU,

project number PNII - IDEI 1238/2008 and by the

Sectoral Operational Programme Human Resources

Development 2007-2013 of the Romanian Ministry

of Labour, Family and Social Protection through the

Financial Agreement POSDRU/6/1.5/S/16.

REFERENCES

Antoniou, G. and van Harmelen, F. (2008). A Semantic

Web Primer 2

edition, MIT Press, USA.

Cernian, A., Bertesteanu, A. et al. (2011), Using Natural

Language to Query an Automatic Control Disciplines’

Ontology, The 18th International Conference on

Control Systems and Computer Science, Bucharest,

Romania.

Gruber, T. R. (1993). Toward principles for the design of

ontologies used for knowledge sharing. In Interna-

tional Journal of Human-Computer Studies, pp. 907–

928.

Horridge, M. (2009). A Practical Guide To Building OWL

Ontologies Using Protégé 4 and CO-ODE Tools,

Edition 1.2, The University of Manchester.

Horridge, M. and Bechhofer, S. (2009). The OWL API: A

Java API for Working with OWL 2 Ontologies. In

OWLED 2009, 6th OWL Experienced and Directions

Workshop, Chantilly, Virginia.

Motik, B., Shearer, R. And Horrocks I. (2009).

Hypertableau Reasoning for Description Logics.

Journal of Artificial Intelligence Research, vol 36, pp.

165-228.

Sierra, K. and Bates, B. (2008). Head First Java 2

Edition, O’Reilly.

Yildiz, B. (2006). Ontology Evolution and Versioning. The

state of the art, Vienna University of Technology,

Institute of Software Technology & Interactive

Systems (ISIS).

Pellet Reasoner: http://clarkparsia.com/pellet

UML Standard: http://www.uml.org/

ICSOFT 2011 - 6th International Conference on Software and Data Technologies

426