INTERACTION PROCESSING OF A COGNITIVE

EDUCATIONAL SYSTEM

Arturo Caravantes

Institute of Educational Sciences, Universidad Politécnica de Madrid, c/Profesor Aranguren s/n, Madrid, Spain

Keywords: Intelligent educational systems, User interface, Cognitive tutoring.

Abstract: This paper briefly describes some of the modelling work of intelligent web-based educational systems with

cognitive monitoring. Specifically, it focuses on the interaction process consisting of a processor based on

mental models and several interaction environments related to interface devices. Both systems work

together to regulate the student learning process through controlling working memory load, focusing

attention and presenting knowledge in conceptual-semantic format. These features require an object-

oriented rich web interface.

1 INTRODUCTION

User interfaces of educational systems have

followed a parallel development to the information

and communication technologies improving their

capability of interaction and presentation. The old

interfaces with command-line textual interaction

were replaced by GUI/WUI (Graphical User

Interfaces / Web User Interfaces) with hypermedia

interaction. The complexity of new requirements for

higher interaction, adaptability and student

monitoring has changed the interface designs from

traditional Web browser to the more sophisticated

Rich Internet Applications (RIA). These interface

applications enable the deployment of new features

like conversational agents (Rus & Graesser, 2006;

Tarau & Figa, 2004; Cassell et al., 2000), 3D

navigational environments and physical agents,

which reduce the ambiguity of linguistic

communication through additional emotional

information (Zakharov et al., 2008; Forbes-Riley et

al., 2008; Prendinger & Ishizuka, 2005).

Diversification of media devices (Internet,

Mobile, PDA) and content globalization has

promoted integration initiatives as the Edit@ project

(www.proyectoedita.org), mainly devoted to the

creation of specifications for resources and user

synchronization such as: SMIL (Synchronized

Multimedia Integration Language), AAIML

(Alternate Abstract Interface Markup Language),

AUIML (Abstract User Interface Markup

Language), UIMLA (User Interface Markup

Language) , XIML (eXtensible Interface extensible

Markup Language), Swing and XUL (XML-based

User-interface Language).

This paper presents a generic interaction

architecture in the context of intelligent web-based

educational systems, especially those focused on

cognitive control (Arrabales & Sanchis, 2008;

Chong et al., 2007; Pinker, 2007; Lehman et al.,

2006; Huss et al., 2006; Bach, 2003; Anderson &

Corbett, 1997). The model must be scalable,

interoperable and designed to monitor mental

processes involved in the teaching-learning process.

This work is part of an ontology modelling project

called COES (Cognitive Ontology of Educational

Systems), implemented on two different intelligent

educational systems: TIX and MAP. The first

approach is a traditional adaptive web-based

educational system while the second one is an

adaptive rich-interface educational system with

cognitive tracking. The TIX system is used to obtain

baseline data to be compared with that from the

MAP system.

In the following section, the COES model of

interaction is introduced. Sections 3 and 4 describe

the main components of the interaction domain in

the MAP architecture: a mental processor and a

conceptual interface.

387

Caravantes A. (2010).

INTERACTION PROCESSING OF A COGNITIVE EDUCATIONAL SYSTEM.

In Proceedings of the 2nd International Conference on Computer Supported Education, pages 387-390

DOI: 10.5220/0002771603870390

 SciTePress

2 INTERACTION MODEL

The traditional Intelligent Educational Systems

(IES) architecture keeps a functional division in

which a processor uses the rules of a

pedagogical/adaptive model to select content from a

domain model depending on a student model. In

COES proposal we have opted for architecture

composed of three functional domains as shown in

Figure 1. The educational domain encodes

knowledge involved in the teaching-learning

process. The personal domain estimates the

characteristics and state of the student, and simulates

his behaviour. The interaction domain, subject of

this paper, facilitates communication with the user

and includes the user interface and the control unit.

Dynamics of web-base educational systems

requires dividing the interaction domain into two

subsystems (Figure 1): an environment that interacts

on sensory level with the user (IE-Interaction

Environment) and another subsystem (IP-Interaction

Processor) that processes all communication with

educational and personal domains, and regulates the

interaction with the user applying perception laws

and cognitive control. Typically, IE and IP run on

separate computers linked by some kind of logical

connection. This functional division allows having a

single processor (IP) for multiple environments (IE)

and thus adapting the teaching-learning process to

the physical display constraints and allowing

ubiquitous-learning.

Figure 1: Scheme of COES architecture focused on the

interaction domain.

Wide system communication takes place via an

action-based protocol. Depending on the direction of

the communication we have two interaction types:

Activators come from the environment and are

generated by the user behaviour and Actuators

generated as responses by the processor.

The processor (IP) controls system running along

all domains and performs the following functions:

origin detection and validation, execution control by

priorities, response monitoring and process-time

control.

3 MENTAL PROCESSOR

The processor (IP) of the MAP system is composed

of interconnected units called nucleus working in

parallel. These cells set an architecture based on

mental processing to enable the learning monitoring

by simulation. An interaction response is made by a

3-stage sequential process laid out by the following

subsystems (Figure 2):

– Sensory System. It processes the inputs from the

environment using 3 nuclei: Origin, Trace and

Message. The first identifies the user and focuses

the attention of the system. The second processes

all actions in the environment to refresh the

memory of remote monitoring. Finally, the

message nucleus transforms student

communications into internal signals.

– Processor System. It keeps track of pending

processing internal signals of the working

memory called gaps, and controls its processing

priorities. It is composed of three main nuclei:

Attention, Memory and Control. The first

focuses and fills internal signals in the working

memory. The second accesses, updates and

reinforces the content of the working memory.

Finally, the control nucleus regulates execution

through the selection of pending gaps. During

processing, these nuclei activate other ones in

complementary subsystems like cognitive,

perceptual, emotional, and behavioural.

– Motor System. It converts logical actuators

generated by the processor system into physical

actuators and motor actions that are sent to the

environment. This conversion is based on

parameters of cognitive load and attention

capability of the user.

CSEDU 2010 - 2nd International Conference on Computer Supported Education

388

Figure 2: Functional scheme of interaction processor.

4 CONCEPTUAL INTERFACE

The interaction environment of the MAP system is a

RWI/OOUI (Rich Web Interface / Object-Oriented

User Interface) application, focused on content

presentation of multichannel conceptual format

(text, image and audio). The MAP system interacts

with students by sending motor actions that

dynamically change environment elements called

SIC (Structural Interaction Components). The

environment synchronizes and executes motor

actions on the client computer, runs the SICs,

controls remote system connections and logs actions

and reactions.

Designers and developers can design new SICs

adapted to the client device. They should always be

derived from one of the 5 generic SIC types

(samples in Figure 3):

– Dialog. Interpellation component (e.g. Message,

Identification, Query…)

– Stimulus. Sensory component (e.g. Background,

Text, Image...)

– Entity. Abstract component (e.g. Icon, Object ...)

– Relation. Association component (e.g.

Implication, Succession, Group ...).

– Stage. Conceptual component that serves to

focus attention.

SIC components have the following common

characteristics:

– Configurable aspect through parameters.

– Scalable sizing and positioning through

percentage values.

– Auto-inclusion. A SIC can contain several SICs.

– Transformable through transitions like moving,

scaling, rotation and flash.

– Active links than include communication options

for the user. Each SIC usually has an

intercommunication button that appropriately

reacts to the mouse-over event.

– Internal control of perception time. Learning

monitoring systems need to know when the

student perceives a SIC.

– Public state for synchronization consisting of

presentation (Permanent/Transient), perception

(Present/Past), blockade (Yes/No), duration

(Initiated/Ended), sound (Active/Stop).

Figure 3: Samples of SIC components.

5 DISCUSSION

This article briefly describes an interaction

architecture focused on cognitive tracking, called

MAP.

The modelling approach was evaluated through a

course about Web Design. 32 students participated in

the study, 17 of them used the TIX system and the

reminder used the MAP system. Academic

performance (P

) and time spent (T

) are dependent

variables of the study. P

is obtained from students’

responses to a final questionnaire ranging 0-100. T

is estimated by the time students remain logged into

the system.

Figure 4: Comparative of students’ performances.

The initial analysis of the results (Figure 4) show

similar academic performance between groups,

lightly higher among users of the MAP system.

Time spent by students in the TIX system exhibits

high variability and a lower average than that

INTERACTION PROCESSING OF A COGNITIVE EDUCATIONAL SYSTEM

389

recorded in the MAP system. This fact is justified

because the TIX interface focuses on textual

contents. Many students prefer to print documents

and read them offline thus reducing logged time.

Further analysis has begun on the student

satisfaction degree and the level of knowledge

acquisition.

REFERENCES

Anderson, J.R. & Corbett, A.T., 1997. Intelligent tutoring

systems. In Handbook of Human-Computer

Interaction. Elsevier Science B.V. pp.849-74.

Arrabales, R. & Sanchis, A., 2008. Applying machine

consciousness models in autonomous situated agents.

Pattern Recognition Letters, 29, pp.1033-38.

Bach, J., 2003. The MicroPsi Agent Architecture. In

Proceedings of the 5th International Conference on

Cognitive Modeling (ICCCM-5). Bamberg, Germany,

2003.

Cassell, J., Sullivan, J. & Prevost, S.a.C.E.F., 2000.

Embodied Conversational Agents. Cambridge,

Massachusetts: The MIT Press.

Chong, H.Q., Tan, A.H. & Ng, G.W., 2007. Integrated

cognitive architectures: a survey. Artificial

Intelligence Review, 28(2), pp.103-30.

Forbes-Riley, K., Rotaru, M. & Litman, D.J., 2008. The

relative impact of student affect on performance

models in a spoken dialogue tutoring system. User

Modeling and User-Adapted Interaction, 18, pp.11-43.

Huss, D., Taatgen, N.A. & Anderson, J.R., 2006.

Instructional Representation in cognitive modeling. In

Proceedings of the Seventh International Conference

on Cognitive Modeling., 2006.

Lehman, J., Laird, J. & Rosenbloom, P., 2006. A gentle

introduction to soar, an architcture for human

cognition: 2006 update. [Online] Available at:

http://ai.eecs.umich.edu/soar/sitemaker/docs/misc/Gen

tleIntroduction-2006.pdf [Accessed 20 december

2009].

Pinker, S., 2007. Cómo funciona la mente. Barcelona:

Ediciones Destino, Colección Imago Mundi.

Prendinger, H. & Ishizuka, M., 2005. The empathic

companion: a character-based interface that addresses

users’ affective states. Applied Artificial Intelligence,

19(3/4), p.267–285.

Rus, V. & Graesser, A.C., 2006. Deeper natural language

processing for evaluating student answers in

intelligent tutoring systems. In American Association

of Artificial Intelligence. Menlo Park, CA, 2006.

AAAI.

Tarau, P. & Figa, E., 2004. Knowledge-Based

Conversational Agents and Virtual Storytelling. In

ACM Symposium on Applied Computing 2004. New

York, 2004.

Zakharov, K., Mitrovic, A. & Johnston, L., 2008. Towards

Emotionally-Intelligent Pedagogical Agents. In ITS

2008., 2008.

CSEDU 2010 - 2nd International Conference on Computer Supported Education

390