In this paper, the concept of collaborative intelligent educational system is presented. Different knowledge representation models are compared in the context of their use in collaborative intelligent educational systems. Advantages of semantic networks for knowledge representation in such systems are described. M ain advantages of extended semantic networks are shown and a set of basic operations regarding them is drawn up.
In today's world, a shift fro m technology based on energy investing to technologies based on knowledge and informat ion has taken place. Knowledge p lays a special role in postindustrial evolution. Leading e xperts in the fie ld of pract ical imp le mentation of the basic principles of sustainable development strategy – Peter Drucker [1], Alvin Toffle r [2], James Brian Qu inn [3] – independently proclaimed hu man entry into a new economy society – knowledge society, in which knowledge is the basic economic resource. Thus, regeneration of this resource has great importance, which is impossible without developing a conception of representation, acquisition, analysis and use of knowledge.
On the other hand, contemporary society faces new challenges – how to organize educational process in such a way that graduates become so called knowledge workers in the full sense of this term [4]?
The purpose of this artic le is to choose a possible model for knowledge representation in intelligent collaborative educational system as a basis for preparation of knowledge workers in today's society.
The leading social groups of the knowledge society will be 'knowledge workers' – knowledge e xecutives who know how to allocate knowledge to productive use [5]. Knowledge wo rk inc ludes the work of a ll partic ipants of the production process to achieve optima l results through combination of each employee’s individual skills. A knowledge worke r must constantly learn innovative knowledge to be co mpetitive, and on the other hand – he or she cannot do without teaching. Thus, emp loyee training function is one of priorities of staff manage ment functions in modern organizat ion s. But in global virtual organizations employees are scattered around the world.
Changes that occur each time and co mp le xity of e lectronic technology society that uses a new type of electronic co mmunicat ion devices have resulted in continuing growth in the amount, diversity and service activities carried out in all fie lds [6]. Thus, in today's society, the concept of collaborative systems has emerged.
A collaborative system is a system in which many users or agents are engaged in a shared activity, often in distanced locations. In the large fa mily of d istributed applications, collaborative systems are distinguished by the fact that the agents work together to achieve a co mmon goal and have a great need to interact with each other in the sense that they share information, change requests, etc. [7]
An educational system is a set of interre lated educational and innovative processes and the manage ment of these processes . There e xist general laws of develop ment of educational systems. These are: focus on the ultimate objective, development of measures for precise system functioning, compliance of the sub-objectives with the ultimate objective, availability of resources, consistency, and safety.
Collaborative systems are applied in the educational fie ld and a imed at evaluating and enhancing the performance of educational p rocess [8]. Collaborative learning he lps knowledge workers to carry on deeper conversations, create mult iple perspectives and develop reliable a rgu ments. This is the main reason why collaborative groups facilitate greater cognitive develop ment than the one that the same individuals can ach ieve while studying alone [9].
An intelligent collaborative educational system applies methods of artific ia l intelligence to provide better support for the users of educational systems and is based on three ele ments : interconnection (a resource sharing technology education), instrumentation (accumu lation of necessary data) and intelligence (making decisions that enhance the learning process) [10]. The architecture of an intelligent collaborative educational system can be described as is shown on Fig. 1. User Module effects interface between the system and the user. Management Module collects informat ion fro m other modules, analyses and processes it, supplies other modules with info rmation obtained fro m analysis. Do ma in knowledge module manages a number of educational objects and provides users with appropriate objects. Collaborative Educational Module pro cures strategies in accordance with the co mmon goal o f education. Control Module affords user tasks and tests and verifies their execution according to its model.
Collaborative Educational
Module Domain Knowledge
Module
User Module Control Module Management Module
Many research articles in the world and in Latvia are dedicated to the problem o f intelligent collaborative educational system development. Though the idea of “intelligence” has built a long tradition in learning systems, it has only relatively recently entered the domain of collaborative learning.
For e xa mple , in [11] it is e mphasized that co mputer-supported collaborative learning is focused on how collaborative lea rning supported by technology can enhance peer interaction and work in groups, and how collaboration and technology facilitate sharing and distribution of knowledge and expertise among community members.
In [12] it is defined that an intelligent educational system a ims to provide learner tailored support during the problem-solving process, as a human tutor would do. The comparative ana lysis of a large a mount of publications resulted in a c lassification scheme , which is proposed as a comp rehensive tool for analy zing and interconnecting the majo r design principles applied in the area of intelligent systems for collaborative learning support. Intelligent systems for collaborative learning support c an be classified and their design and operation can be analyzed in the following dimensions: • Pedagogical objective: the general pedagogical objective of the system. • Target of intervention: the focus of the intelligent support. • Modelling: modelling techniques implemented in the system. • Technology: the kind of technology used to imple ment the intelligent operation of the system. • Design space: how the intelligence-based intervention is presented to the partners.
Similar proble ms are described in many artic les : development of intelligent tutoring
systems based on knowledge workers ’ personal knowledge of management systems [4,
13], use of concept maps in adaptive and intelligent knowledge assessment [14-16],
mu ltiagent techniques, development of agent based intelligent tutoring systems [17-18],
possibilit ies of e xtending the agent oriented software engineering methodology to make it
usable for the development of other agent oriented systems [19]; develop ment of
webbased collaborative e-learn ing systems [20]. Paper [21] presents a method for agents’
knowledge representation by using semantic network; paper [
However, it should be noted that the problem of develop ment of direct ly collaborative educational systems for knowledge worke rs considering a set of their attributes has not been paid the attention it deserves in the scientific community. 4
Solving the p roble m of knowledge representation in genera l involves ensuring adequate
display of knowledge stored in a variety of knowledge sources as formalized
representations, automated processing of which will allow to effective ly solve proble m
tasks from the do main under research to obtain results that are not qualitatively infe rio r to
results obtained by highly qualified specialists of this domain in dealing with such
problems [
At present, the most common models of knowledge representation in intelligent systems are the following: • logical models (knowledge is represented as a set of correctly formed formulas o f any formal system); • production models (central e le ment of th is model is a set of products and rules of inference, instead of logica l infe rence characteristic of logical models the conclusion based on knowledge is used); • fra me -based models (foundation of this knowledge model is the concept of fra me – a data structure that represents some conceptual object or standard situation ); • network mode ls in which the do ma in is considered as a set of objects and their relationships. Semantic networks are the most common network model of knowledge representation.
In the applied systems of artific ia l intelligence in fields like medic ine, b ioinformatics,
semantic web etc., ontologies are widely used for knowledge representation. In working
with ontologies , special languages of ontology representation (RDFS, OW L etc., [
Based on the set of general require ments traditionally imposed on models, it is
possible to formulate some require ments for models of knowledge representation in
intelligent collaborative educational systems [
Let us consider the comparison of properties of the basic mode ls of knowledge
representation based on the aforementioned requirements.
The data presented in Table 1 ("+" symbol indicates the presence of corresponding
properties in the representation model, “” symbol – partia l presence, “–” symbol –
absence) shows that semantic network model meets the greatest number of require ments.
As a result, the model of knowledge representation to be developed should be based on
this model taking into account development of its logica l and procedural properties.
Extended semantic network is one of the varieties of this improved semantic network
models. Extended semantic netwo rks have been developed to eliminate heterogeneity of
the usual semantic networks, which is caused by presence of the aggregates of objects
lin ked by t ies of relat ions, interconnected relations and other factors [
The concept of semantic networks of knowledge repres entation is based on the idea that all knowledge can be represented as a set of objects (concepts) and links (relations) between them.
The main advantages of semantic networks are as follows: • proximity of the network structure to the semantic structure of phrases in natural language; • visuality of knowledge systemrepresented graphically; • versatility achieved by selecting an appropriate set of relations; • possibility to connect different network fragments; • definition of operations performed on objects; • for each operation on data or knowledge, it is possible to allocate a certain part of network that covers the essential characteristics of request. • Semantic models also have some disadvantages : • an arbitra ry structure and different node and relat ion types complicate informat ion processing; • semantic networks have no special means to determine time dependencies; • complexity of exception handling; • representation, us age, and modification of knowledge and description of systems at the level of complexity corresponding to that of reality is a time-consuming procedure; • processing of network mode ls requires a special apparatus of forma l withdrawal and planning.
In e xtended semantic networks , nodes can correspond not only to objects or concepts, but
also to relations, logica l co mponents of information, co mp le x objects and others . To
everything that can be regarded as an independent unit , its own node must correspond.
Thus, nodes of different type are entered – nodes corresponding to names of relat ions, as
well as a special co mposite ele ment called connection node. They are connected by
ma rked edges with nodes taken from the array of above-mentioned nodes. As a result, a
frag ment appears that corresponds to elementary situation, i.e. objects that are bound by
relation. Such a fragment is called an elementary one [
The basis of e xtended semantic networks is a set of nodes fro m which e le mentary
frag ments D0 (D1, D2, ... , Dk / Dk+1) are co mp iled, where D0 stands for re lation na me; D1,
D2, ... , Dk – for the objects participating in relation; Dk+1 – for the connection node
denoting the whole array of objects participating in the relat ion; this node is also called c
node of elementary fragment; D0, D1, D2, ... , Dk+1 D, к>0 [
Extended semantic networks are considered as a finite set of elementary frag ments . Na mes of relat ions play the role of objects and can enter into relations . This defines high homogeneity of the model. Connection node of elementary frag ment can b e part of other elementary fragments but not as a c-node.
The set of nodes is partitioned into three disjoint subsets [
D = G∪X∪E , where G stands for detected nodes (definite co mponents ); X – for undetected nodes (variable co mponents, their roles are defined in further processing of the model; Е – fo r special nodes (used in the production description).
In its turn, the G set consists of three subsets :
G = R∪A∪{t,f} , where R stands for an array of re lation na mes (corresponding to D0); А – for an array of concepts; {t,f} – for logical co mponent D pointing to the truth or fa lsity of the re lations of the represented relation, t – true, f – false.
Let us represent, for e xa mp le, the most common vie w of intelligent collaborative educational system in the form of a frag ment of e xtended semantic network. We shall use a reduced image of the e le mentary frag ment. At first, we don’t draw the node {t,f} with its edges. We assume that if this node is not connected to connection node by dint of corresponding frag ment, true (t) re lations are represented. It will simp lify the drawing . Secondly, we put a special character of the e mpty space (_) in comp liance with unimportant or unessential components . This symbol is required to represent relations for which not all components are specified but only the necessary ones.
As already mentioned, a collaborative intelligent educational system (CIES) is based on four ele ments : interconnection (IC), instrumentation (INST), intelligence (INT) and shared activity (SA). Thus, we can write the following expression:
Operations on elementary frag ments can be expressed by the objects themselves. The
following nodes are entered into R to represent operations [
For processing of knowledge represented by extended semantic networks , the princip le o f
matching pattern in the form of t wo network overlay method is used. It is based on
identification of rules a llo wing to b ind nodes and compare networks in accordance with
laws of logic [
The authors think that it is necessary to modify a set of operations on ele mentary frag ments to use extended semantic networks for solving the proble m of collaborative intelligent educational system development, because the whole domain of logica l connections cannot be represented in e xisting e xtended semantic network mode ls. The future work of the authors will be devoted to this issue. 6
In this paper, concepts of knowledge worker and collaborative intelligent educational systems have been considered, basic models of knowledge representation have been reviewed. The authors have briefly described advantages of e xtended semantic networks for knowledge representation in these systems and have shown that these networks meet the requirements more fully. Thus, the authors think that use of extended semantic networks in their future work is e xpedient. Future work will focus on the evolution of e xtended semantic network models, the general scheme of the imp rovement and detailed elaboration of collaborative intelligent educational systems, their p ractical imple mentation as a prototype and approbation in real conditions of modern higher education institution s. 1. 2. 3. on Business Informatics Research: International Workshops and Doctoral Consortium, 2011, p p . 143-157. 5. Drucker P. F. Post-Capitalist Society. Harper Business, USA, 1994. 240 p. 6. Yi-Chen Lan. Global information society: operating information systems in a dynamic global business environment. Idea Group Publishing, 2005. [Online] available at http://books.google.lv/ 7. Dobrican O. An Example of Collaborative System // International Workshop Collaborative
Support Systems in Business and Education, Cluj-Napoca, 2005, pp.48. 8. Ciurea C., Batagan R. Collaborative Educational Systems // Oeconomics of Knowledge,
Volume 2, Issue 4, 4Q 2010, pp.13-22 9. Różewski P., Jankowskia J., Bródka P., M ichalski R. Knowledge workers’ collaborative learning behavior modeling in an organizational social network. Computers in Human Behavior, Volume 51, Part B, October 2015, p p . 1248–1260. 10. Batagan L., Boja C., Cristian I. Intelligent Educational Systems, Support for an Education
Cluster // Proceedings of the 5th conference on European computing conference, pp. 468-473 11. Lipponen L. Exploring Foundations for Computer-Supported Collaborative Learning // Proceedings of the Fourth Computer Support for Collaborative Learning Conference: Foundations for a CSCL Community, 2002, pp.72-81 12. M agnisalis I., Demetriadis S., Karakostas A. Adaptive and Intelligent Systems for Collaborative Learning Support: A Review of the Field // IEEE Transaction on Learning Technologies, Vol. 4, No. 1, January -M arch 2011, pp. 5-20 13. Grundspenkis, J., Ciekure, J. The Conceptual Framework for Integration of Intelligent Tutoring and Knowledge M anagement Systems in Education Settings // Perspectives in Business Informatics Research: 10th International Conference, 2010, pp. 242-250. 14. Grundspenkis, J. Concept M ap Based Intelligent Knowledge Assessment System: Experience of Development and Practical Use. No: M ultiple Perspectives on Problem Solving and Learning in the Digital A ge. New York, Dordrecht, Heidelberg, London: Springer Science+Business M edia, LLC, 2011. pp. 179-198. 15. Anohina-Naumeca, A., Grundspenkis, J. Concept M aps as a Tool for Extended Support of Intelligent Knowledge Assessment // Proceedings of the 5th Internat ional Conference on Concept M apping: 5th International Conference on Concept M apping, 2012, pp. 57-60. 16. Anohina-Naumeca, A., Graudin a, V., Grundspenkis, J. Using Concept M aps in Adaptive Knowledge Assessment // Advances in Information Systems Development: N ew M ethods and Practice for the Networked Society. Vol.1. New York: Springer, 2007, pp. 469-480. 17. Lavendelis, E., Grundspenkis, J. MIPITS - An Agent based Intelligent Tutoring System //
Proceedings of 2nd International Conference on Agents and Artificial Intelligence, 2010, pp. 5-13. 18. Lavendelis, E., Grundspenkis, J. MASITS M ethodology Supported Development of Agent Based Intelligent Tutoring System M IPITS // Proceedings of 2nd International Conference on Agents and Artificial Intelligence, 2010, pp. 119-132. 19. Lavendelis, E. Extending the M ASITS M ethodology for General Purpose Agent Oriented Software Engineering // Proceedings of the 7th International Conference on A gents and Artificial Intelligence, 2015, pp. 157-165. 20. Novickis, L., Rikure, T. Intelligent Tutoring Systems // Proceedings of IST4Balt International Workshop “IST 6th Framework Programme – Great Opportunity for Cooperation & Collaboration”, 2005, pp. 35-40. 21. Hernes M . The Semantic M ethod for Agents’ Knowledge Representation in the Cognitive Integrated M anagement Information System // Position Papers of the Federated Conference on Computer Science and Information Systems, 2015, Vol. 6, pp. 195–202