74 Economy Informatics, vol. 10, no.

1/2010

Learning Intelligent Collaborative Systems

Loredana MOCEAN, Robert Andrei BUCHMANN, Monica CIACA
Babes – Bolyai University of Cluj – Napoca, Romania
Faculty of Economics and Business Administration
Business Information Systems Department
{loredana.mocean, robert.buchmann, monica.ciaca}@econ.ubbcluj.ro

This paper proposes a collaborative learning model based on a semantic module detecting
concepts that were not properly acquired during the learning process. A database structured
is proposed which was designed based on the on-line collaborative learning and social
networking requirements. The objective of the research is to implement an intelligent and
flexible on-line intelligent collaborative learning system and to facilitate students in
increasing their performance within on-line learning.
Keywords: Learning, On-line Learning, Intelligent Learning, Database, Semantic Web

1 Introduction
The Information Society has been
developing as a new step in our social
content production. The development of new
learning and sharing methods represents an
essential factor for increasing
evolution, by intensively exploiting competitiveness, modernizing services and
information in acquisition and exchange of developing new ways of communication
all types of resources. The technological between individuals. From the perspective of
support on which the new ways of chain consisting in data-information-
knowledge assimilation are based is knowledge, the modern learning processes
determined by the convergence of three can be classified in four categories, as shown
fields: information technology, in figure 1.
communication technology and digital

Fig. 1. The evolution of learning paradigm

In time, information systems and services for

assisting learning followed various 2 Problem context
methodologies and models, in order to The learning paradigm in the modern
improve management effectiveness and organization is not limited to a process of
efficiency of learning resources. iterative understanding of a limited set of
The paper is structured in several sections knowledge. New knowledge emerges or is
presenting the problem context, the proposed produced frequently from various sources, be
model and a SWOT analysis. it direct acquisition or inferring rules.
Economy Informatics, vol. 10, no. 1/2010 75

Knowledge can be grouped in two The new dimensions of learning in modern

categories: individual and organizational society are described in figure 2.
(corporate).

Fig. 2. New dimensions and types of learning in modern society

Collaborative learning is a method of and source (home, school, workplace or any

teaching and learning in which student’s other place where the individual engages in
team work together to explore a significant interactions with other members of the same
question or create a meaningful project. A group). Informal learning is part of the daily
group of students discussing a lecture or work and routine, influenced by external or
students from different schools working internal impulses and strongly related by
together over the Internet on a shared informal knowledge exchange ([2]).
assignment are both examples of Nowadays, as computer-based
collaborative learning [1]. communication invades this routine and
Cooperative learning is a specific kind of becomes a way of life, informal learning
collaborative learning. In cooperative tends to integrate computer-based processes.
learning, students work together in small Informal learning has several features:
groups on a structured activity. They are  is mainly inductive;
individually accountable for their work, and  is stimulated when there's a need for
the work of the group as a whole is also increased learning speed;
assessed. Cooperative groups work face-to-  does not emphasize knowledge accuracy
face and learn to work as a team [1]. ([2]), but rather the diversity of sources
The formal learning is learning that takes and their reputation;
place within a teacher-student relationship,  the knowledge acquisition is not
such as preschool learning, school learning, necessarily explicit and formalized, but
high school or university. knowledge is often applied real-time on a
The non-formal learning is organized pragmatic level;
learning outside the formal learning system.  is associated with implicit learning and
For example: learning by coming together tacit knowledge ([3])
with people with similar interests and Methods of informal learning
exchanging viewpoints, in clubs or in Merriam et al. ([4]) synthesize three ways of
(international) youth organizations, learning, which differ based on intentionality,
workshops. awareness and time, as follows:
The informal learning is the category that • self-directed learning, using the computer
implies nonsystemic learning through as a tool;
everyday experience, regardless of context • incidental learning, using the computer as
76 Economy Informatics, vol. 10, no. 1/2010

an assistant; basic RDF assertion semantics using a

• socialization and tacit learning, using the relational-knowledge mapping system that
computer as an environment. clearly delimits acquired and unacquired
knowledge concepts for each student. This
3 Related works provides input for a potential
Several models for intelligent e-learning recommendation system by extracting the
systems have been proposed throughout the status of learning concepts from RDF graphs.
literature. In [5] the authors propose a formal The collaborative aspect is defined by the
concept analysis methodology for developing teacher’s involvement in modeling the
learning models based on an a priori concept graphs, with respect to the
knowledge corpus. [6] provides a assessment of student knowledge for a
comprehensive overview of the evolution of certain knowledge context.
metadata in e-learning applications from Collaborative learning is, first of all, a
standards to specialized representations. One philosophy of interaction and lifestyle. More
of the closest related approach is [7], where specifically, it designates a methodology of
students mistakes are exploited in an e- learning and a certain interaction structure
learning recommender environment in order which tends to follow a common goal.
to control acquired knowledge within a The collaborative learning mechanisms, as
learning agents framework. Compared to described by Kegan in 1998 [8] are reflected
these studies, our approach is grounded on in figure 3.

Fig. 3. Collaborative learning mechanisms

Even before 1960, before personal computers exercise.

changed the education paradigm, researchers b. Intelligent Tutoring Systems
investigated the effectiveness of Features:
collaborative / cooperative learning.  cognitive approach;
Koschmann [9] identified several  learning means mental models;
approaches, which integrate traditional  teachings is a process of model
teaching methodologies with computerized generation and description.
support: c. Logo
a. Computer Assisted Instruction Features:
Features:  constructivist approach;
 behaviorist approach;  pupils build their own knowledge;
 learning means memorizing facts;  teaching is a process of defining a
 the domain knowledge is decomposed in stimulative environment which can be
atomic facts exposed to pupils in logical explored and discovered through
sequence, through instructions and observation and reasoning.
Economy Informatics, vol. 10, no. 1/2010 77

d. Computer Supported Collaborative artificial intelligence because it emulates

Learning actions of a human mediator, providing
Features: answers to pupil input, analyzing problem
 pupils communicate and collaborate; solving strategies and comparing pupil
 pupils are organized in learning actions with pre-programmed models of
communities of various structures and correct and erroneous understanding. The
granularities [10] synergy between intelligent systems and
Intelligent collaborative learning systems CSCL is illustrated in figure 4.
are a CSCL example from the area of

Fig. 4. The synergy between intelligent systems and CSCL

The computer becomes a cognitive tool for collaborative paradigm in a learning science
social-driven learning. CSCL emerged as a context.
reaction to previous attempts of involving The computer's role in the learning process is
technology in education and previous illustrated in figure 5.
approaches for understanding the

Fig. 5. The computer's role in training

4 The proposed model implementation details for a proposed model

This section is dedicated to describing the of collaborative learning systems.
architectural solution and some design and
78 Economy Informatics, vol. 10, no. 1/2010

Fig. 6. The general architectural solution

As the general; architecture diagram shows, a the teacher’s underlying topic-level ontology
classic e-learning system is extended with a and the student concept-level evaluation.
concept management system based on the After an initial login procedure, the user
RDF data model and related querying layers interface elements from figure 7 would be
and wrappers. A learning concept network is provided to the teacher:
defined by merging graphs defined both by

Fig. 7. The teacher’s user interface

The teacher user interface specification is presented in figure 8.

Economy Informatics, vol. 10, no. 1/2010 79

Fig. 8. The Course Window

The student form specification is modeled as follows in figure 9.

Fig. 9. The Student’s Window

Database structure Department, Title, E-Mail Adress)

The model is driven by a relational database STUDENT (IDStudent, Name, Class,
with 11 tabels. The structure of these tabels RegsitrationNo, Birth_Date, E_Mail Adress)
are: SESSION (IDProfessor, Begin_date,
End_Date, End_Time, Room)
ACCOUNTS (E-MAIL Adress, Type, PROF_COURSE(IDProfessor, IDCourse)
Password) STUD_COURSE (IDStudent, IDCourse)
PROFESSOR (IDProfessor, Name, COURSE (IDCourse, Course, Type)
80 Economy Informatics, vol. 10, no. 1/2010

COURSE_FILES (IDFile, Path, IDCourse)  to support the registration and login of

EVALUATION (IDCourse, Evaluation the users;
Type, Questionnaire, Case_Study,  to support searching based on a full text
Laboratory) index and course filtering/browsing;
EXAM_FILES (IDCourse, Path,  to support message exchange
IDEvaluation)  to support rights and privileges for
MESSAGE(IDMessage, IDProfessor, various types of users;
IDStudent, Date, Text)  to support graphical interfaces for various
types of users.
The relationships have been defined in order The tables are linked according to the schema
to meet several requirements: from figure 10.

Fig. 10. The presentation of the relationships between tables, using entity-relationship
diagram

For example, we made a capture of the relationships between the tables (see figure
window in which are reflected the 11).
Economy Informatics, vol. 10, no. 1/2010 81

Fig. 11. The database relationships

The database is mapped on an RDF domain. In order to work, the courses must
repository describing relationships between be backed up by an RDF concept graph
concepts involved in the evaluation process. repository, stored in a persistent way. The
The evaluation table is based on the usual solution for this is a hybrid data-
assumption that each transaction (question) is knowledge base in which the RDF triple
mapped to one or more concept and a correct structure is mapped to a database structure.
answer validates the student knowledge Most semantic libraries allow this, and also
against those concepts. After an evaluation our technology of choice, Python backed up
session, the student will have covered a with RDFLib[11]. From within the proposed
certain part of the concept graph, while system, the stored graphs will be accessed
incorrect questions will reveal the limit of his and queried through SPARQL queries and
„knowledge domain”. Subsequent training the object-RDF mapping layer provided by
sessions will automatically emphasize and SuRF[12].
recommend the study of concepts out of that

Fig. 12. The concept graph

The concept graph is not to be confused with a feasible to setup repositories or ontologies for
class hierarchy. Instead, it is rather a mapping each subject that is taught during the e-learning
of how concepts reflected by learning objects program. The graph is a representation of the
depend on each other’s understanding in order logical dependencies between the topics and
to support a consistent and coherent serial notions within a certain discipline. It is created
acquisition of the information. Also, the graph by each teacher for each course using a visual
is not a domain ontology as it would not be environment such as IsaViz, which in turn
82 Economy Informatics, vol. 10, no. 1/2010

exports it in a variety of RDF serialization object.attribute=value

formats, N3 being preferred for simplicity.
The above graph would be stored using N3 as As the open world assumption permits it,
follows: after a student evaluation, the concept graph
is automatically expanded with properties
<#concept1> <#dependson> that express if the student acquired a concept
<#concept2> .
<#concept2> <#dependson>
or not, depending on a preset evaluation
<#concept3> . scheme. Using CONSTRUCT queries, the
<#concept2> <#dependson> graph of unacquired concept is extracted and
<#concept4> .
further study on file resources linked (also
Libraries such as RDFLib provide means of: through RDF) to those specific concepts is
proposed during the next sessions.
• parsing such a graph:
import rdflib 5 SWOT Analysis
from rdflib.graph import Strong points:
ConjunctiveGraph
graph=ConjunctiveGraph() The improvement of the human resource
graph.parse(„graph- quality:
file.nt”,format=”nt”)  continuous development of teaching
skills, by assimilating skills related to on-
• storing it in a persistent datastore line training systems and on-line course
store = rdflib.plugin.get('MySQL',
development;
Store)('mystore')  teachers can, in turn, promote and
a = integrate their on-line experience with
store.open("host=localhost,passwor
d=admin,user=admin,db=mystore", other projects and teams where they are
create=True) involved;
graph =  higher efficiency and, if carefully
rdflib.ConjunctiveGraph(store)
managed, effectiveness of the educational
process;
• querying it by triple matching or by
 specific experience gain regarding on-
SPARQL queries; new graphs can be
line evaluation methods: questionnaire,
built, seralized and stored if the
multiple choice test, student project;
SPARQL query results in a new
Weak points:
graph:
 resistance of some teachers against
results = graph.query(‘ CONSTRUCT modern learning technology;
?conca <#dependson> ?concb  weak involvement of some teachers due
WHERE { to relaxed terms of usage;
<#concept1> <#dependson>
?conca .  weak involvement of some students due
} ‘) to the lack of self-motivation;
 the cost of technology;
The object-RDF mapping layer provided by  possible software glitches or usability
the SuRF library facilitates the manipulation issues, with no human assistant to rely on
of RDF triples in an objectual syntax: in a real-time manner discouraging the
users;
<#subject><#predicate><#object>  the variety among starting skills for
translates to students, which raises a specific
<#object><#attribute><#value> and can requirement of adaptively.
be expressed in the object-oriented Opportunities:
qualification syntax of the host language,  e-learning improves auxiliary skills both
Python in our case: for trainers and pupils, related to the use
of technology; the experience and the
Economy Informatics, vol. 10, no. 1/2010 83

skills will prove helpful in other contexts; project CONTO PN II 91-037/2007 financed
 e-learning organizations can easily join a through the National Research Authority,
great diversity of European projects managed by Prof. Dr. Nitchi Stefan.
focused on e-inclusion and more efficient
learning systems; References
 e-learning is in itself an occasion to bring [1] Thirteen Online, Cooperative and
together specialists for various fields and Collaborative Learning Workshop.
define new methodologies and [Online] Available at:
applications; http://www.thirteen.org/edonline/concept
 e-learning provides higher dynamics in 2class/coopcollab/index.html
professional development by decreasing [2] V. J. Marsick, “Strategic Organizational
the educational time resource Learning: How Innovative Companies
consumption. Use Learning to Address Business
Threats: Challenges,” Linkage Performance 2002
 e-learning and distance learning seems to Conference, Miami, FL, November 2002.
be related to a more superficial approach [3] M. Polanyi, The Tacit Dimension, Anchor
from students who are accustomed to Books, Double Day and Company,
traditional learning and have problems Garden City, New York, 1967.
with defining self-motivation [4] S. B. Merriam et al., “Updating our
mechanisms; knowledge of adult learning”. In Journal
 on-line learning changes the business of Continuing education in the Health
model of education on all its levels and Professions, Vol. 16, No. 3, pp. 136 –
must be carefully managed; 143, 2007.
 e-learning is subject to technological [5] G. Beydoun, “Formal concept analysis
obsolescence and it is highly dynamic for an e-learning semantic web Source”.
from a management perspective; Expert Systems with Applications: An
 there's a need for automated arbitration International Journal. Vol. 36, No. 8,
systems such as [13] for student evaluation October 2009, pp. 10952-10961, 2009.
processes; [6] H. S. Al-Khalifa and H. C. Davis, “The
evolution of metadata from standards to
6 Conclusions semantics in E-learning applications”. In
Our model targets training providers and Proceedings of the seventeenth
consumers who want to overcome spatial conference on hypertext and hypermedia,
limitation. It provides on-line content Odense, Denmark, pp. 69 – 72, 2006,
management, within a collaborative ACM Press, New York, NY, USA
environment based on communication [7] A. Gladun, J. Rogushina, F. Garcia-
channels and a module for semantic Sanchez et al., “An application of
management of taught concepts and topics. intelligent techniques and semantic web
Besides the traditional social learning technologies in e-learning environments,”
elements, the model involves emerging In Expert Systems with Applications: An
technology in order to define a „separation of International Journal, Vol. 36 , No. 2,
concerns”-type of mapping, allowing March 2009, pp. 1922-1931, Pergamon
programmers to manipulate RDF triples in a Press, Inc. Tarrytown, NY, USA
object-oriented manner and allowing teachers [8] R. Kegan, L. Lahey and E. Souvain,
to integrate their own concept graphs, “From Taxonomy to Ontogeny: Thoughts
visually defined through IsaViz, with a on Loevinger's Theory in Relation to
persistent concept graph store through the Subject-Object Psychology”, In P.M.
simple N3 serialization format. Westenberg, A. Blasi, and L.D. Cohn,
Acknowledgment Eds. Personality Development:
This paper presents research supported by Theoretical, Empirical, and Clinical
84 Economy Informatics, vol. 10, no. 1/2010

Investigations of Loevinger's Conception Computing Conference, 2008, pp.211-

of Ego Development, Mahwah, NJ and 217
London: Lawrence Erlbaum Associates, [14] G. Stahl, T. Koschmann and D. Suthers,
1998. “Computer-supported collaborative
[9] T. Koschmann, Paradigm shifts and learning: An historical perspective,” In R.
instructional technology, Mahwah, NJ: K. Sawyer (Ed.), Cambridge handbook of
Lawrence Erlbaum, pp. 1-23, 1996 the learning sciences, pp. 409-426,
[10] S. Trăuşan-Matu, Interacţiunea Cambridge, UK: Cambridge University
conversaţională în sistemele Press, 2006.
collaborative pe Web, Ed. MatrixRom, [15] D. Fotache D., Groupware, Ed. Polirom,
2008. 2002.
[11] The Official RDFLib support site. [16] Gh. Iosif, A. Marhan, I. Juvina and S.
[Online] Available at: http://rdflib.net Trausan–Matu, “Proiectarea cooperativa
[12] The Official SurfRDF mapper support a unui sistem inteligent de instruire pe
site. [Online]. Available at: Web,” Revista Informatica Economica,
http://code.google.com/p/surfrdf/ No. 3 (19), 2001
[13] R. A. Buchmann and S. Jecan, “An [17] L. Mocean and M. Ciaca, “About
Arbitration System for Student Modeling The ERP Systems”, In Studia
Evaluation based on XML Signature,” Universitatis Babes-Bolyai Oeconomica,
Proceedings of the 2nd WSEAS European Vol. 54, No. 1, pp. 78-86

Loredana MOCEAN has graduated Babes-Bolyai University of Cluj-

Napoca, the Faculty of Computer Science in 1993, she holds a PhD diploma
in Economics from 2003 and she had gone through didactic position of
assistant and lecturer, since 2000 when she joined the staff of the Babes-
Bolyai University of Cluj-Napoca, Faculty of Economics and Business
Administration. In 2009 she has graduated the Faculty of Economics and
Business Administration. She is the author of more than 10 books and over
35 journal articles in the field of Databases, Data mining, Web Services, Web Ontology, ERP
Systems and much more.

Robert BUCHMANN has graduated Babes-Bolyai University of Cluj-

Napoca, the Faculty of Economic Sciences and Business Administration in
2000, holds a PhD diploma in Economics from 2005. He has gone through
the positions of Ph.D. student, assistant and lecturer, since 2001 when he
joined the staff of the Business Information Department within the Faculty of
Economic Sciences and Business Administration. He managed 3 research
projects and is the author of 2 books and over 30 scientific papers published
in academic journals or conference proceedings. His fields of interest are SemanticWeb, E-
business and Software Quality.

Monica CIACA has graduated Babes-Bolyai University of Cluj-Napoca, the

Faculty of Computer Science in 1993, she holds a PhD diploma in
Mathematics from 2002 and she had gone through didactic position of
assistant, lecturer and associate professor, since 1994 when she joined the
staff of the Babes-Bolyai University of Cluj-Napoca, Faculty of Economics
and Business Administration. She is the author of more than 10 books and
over 45 journal articles in the field of Databases, Soft-ware Engineering and
Artificial Intelligence.