The peer to peer E-learning

blabbedharborΤεχνίτη Νοημοσύνη και Ρομποτική

23 Φεβ 2014 (πριν από 3 χρόνια και 3 μήνες)

81 εμφανίσεις

Vancouver, Canada, October 24
-
28, 2005.

The peer to peer E
-
learning

using EECA based on the PECS model

Mahmoud NEJI

--

Mohamed BEN AMMAR
--

M.Adel ALIMI

Mahmoud.Neji@fsegs.rnu.tn, {mohamed.benammar,adel.alimi}ieee.org

RE
search
G
roup on
I
ntelligent
M
achine


2

E
-
Learn 2005


Vancouver, Canada,

October 24
-
28, 2005.



Outline


Introduction


Context



Communication in the group



The peer to peer e
-
learning



EECA:

Emotional Embodied Conversational Agent



PECS:
(Physics, Emotion, Cognition, and Social
Status)


EML:

Emotional Markup Language


Conclusion and Future work


3

E
-
Learn 2005


Vancouver, Canada,

October 24
-
28, 2005.



Introduction


The

e
-
learning

is

an

activity

in

which

as

teachers

and

learners

are

dispersed

geographically
.


Consequently,

the

technology

plays

an

important

role

while

facilitating

the

communication,

the

interaction,

as

well

as

the

necessary

tool

provision

to

the

learning

process
.


Teaching

means

observing

the

behaviour

of

the

pupil,

detect

signs

engaging

and

acuity

in

order

to

detect

affective

responses,

which

may

be

the

manifestation

of

feelings

interest,

excitation,

confusion



Several

e
-
learning

software’s

put

in

scene

virtual

beings

which

accompany

the

learner

in

its

learning

process
.

These

assistants

are

used

as

interface

of

communication

by

which

the

system

gives

directives

to

the

learner
.



4

E
-
Learn 2005


Vancouver, Canada,

October 24
-
28, 2005.





We

present

our

system

of

collective

and

collaborative

learning

on

the

network

P
2
P,

while

being

based

on

PECS

(Physics,

Emotion,

Cognition,

Social

status)

model

integrated

in

the

EECA

(emotional

embodied

conversational

agent

)
.



The

basic

idea

of

such

a

system

rests

on

the

advantages

of

technology

peer

to

peer

(P
2
P)

and

the

development

of

added

value

services

according

to

the

emotional

state

of

the

learner
.




In

a

more

general

way,

the

thorough

knowledge

of

the

user

makes

it

possible

the

computer

to

adapt

to

the

circumstances,

to

react

more

intelligently

and

to

even

anticipate

the

needs

for

the

learner
.


Context


5

E
-
Learn 2005


Vancouver, Canada,

October 24
-
28, 2005.



Communication in the group


A

:

Learner

Em:

Emotion

Propagation of the emotion (Emi) and assistance


from learners

Then

learners
:

After

detection

and

recognition

of

emotion

(Emi)

of

learner

(Ai)

in

the

instant

(ti)

by

the

emotional

agents,

the

emotion

will

be

sent

to

the

EECAi

that

propagates

it

to

the

different

other

EECA
.

If

a

learner

can

assist

him,

he

formulates

his

message

of

help

and

the

send

it

with

his

EECA
.



6

E
-
Learn 2005


Vancouver, Canada,

October 24
-
28, 2005.



Propagation of the emotion (Emi) and assistance from Tutor


Then

tutor
:

The

tutor

intervenes

in

the

case

where

the

emotion

in

question

correlates

to

a

difficulty

met

by

most

learners
.

In

this

context

-
under

the

management

of

agent

curriculum
-

he

must

rephrase

question

or

provide

other

explanations
.



7

E
-
Learn 2005


Vancouver, Canada,

October 24
-
28, 2005.



The

Peer

to

Peer

e
-
learning

P2P


Arch.
Centralized / Decentralized

Learner

EECA

Tutor

The P2P is a concept of a communication on a network like the Internet in which a node
can be a server or a client. That's allows to share and to benefit all the resources and the
available services in the system.



8

E
-
Learn 2005


Vancouver, Canada,

October 24
-
28, 2005.



Peer


Architecture

Learner

EECA

Tutor

Peer

Peer

Emotional

System

P
2
P

systems

are

application

level,

collaborative

systems

where

agents

work

together

to

perform

certain

tasks
;

thus

multi
-
agent

system

technology

seems

very

relevant

for

implementing

these

types

of

systems
.


9

E
-
Learn 2005


Vancouver, Canada,

October 24
-
28, 2005.



EECA

:


Emotional

Embodied

Conversational

Agent


Interest

of

EECA


Our

objective

is

to

produce

an

emotional

visual

entity

be

in

charge

by

several

software’s,

able

at

the

same

time

to

enclose,

evaluate

and

react

with

the

effort

of

the

learner,

the

EECA

have

three

main

features

permitting

them

to

answer

to

this

new

need
:




They

are

agents




They

are

interactive




They

aim

the

natural

interaction




They

are

conversational,

according

to

several

modalities

(movements,

word,

expressions,

)



They

are

animated


10

E
-
Learn 2005


Vancouver, Canada,

October 24
-
28, 2005.





Role of EECA


In our system, the EECA can play three main roles :


Assistants


Partners of actors in the virtual environments


Need of tutoring


Behaviour of EECA


Among all behaviours that can be present, we can mention :


Converse


Context


Accompany


Appearance


Intensity: +/
-

theatrical; +/
-

discreet (a virtual agent must
replicate non verbal signals, generally moderate, of a human
agent).



11

E
-
Learn 2005


Vancouver, Canada,

October 24
-
28, 2005.



PECS

:


Physics,

Emotion,

Cognition,

and

Social

Status


Simulate

the

human

behaviour

in

a

group
.



Based

on

the

integration

of

physical,

emotional,

cognitive

and

social

features
.


The

PECS

agent

architecture

was

applied

to

a

group

learning

case

study

as

a

case

where

psychological

and

social

mechanisms

should

be

considered

in

a

group

integration

context
.


Replace

the

BDI

architecture

(Belief,

Desire,

Intention)



12

E
-
Learn 2005


Vancouver, Canada,

October 24
-
28, 2005.



The

PECS

reference

model

opens

up

new

challenging

possibilities

for

the

modeling

of

systems

that

include

human

factors

as

important

and

decisive

subcomponents
.


PECS

is

especially

useful

when

complex

human

behaviour

has

to

be

taken

into

account
.

This

includes

physical

conditions,

emotional

states,

cognitive

capabilities

and

the

social

status

along

with

accompanying

mutual

interactions
.


13

E
-
Learn 2005


Vancouver, Canada,

October 24
-
28, 2005.



PECS

is

therefore

an

agent

architecture

that

follows

those

two

ideas
.

Therefore,

PECS

agent

model

consists

of

three

horizontal

layers
:

1.
Information

Input

Layer
.

This

layer

processes

the

input

taken

from

the

agent

environment

and

consists

of

two

components
:

The

Sensor

and

Perception

components
.


2.
Internal

Components’

Layer
.

The

personality

of

the

agent

is

modelled

at

the

second

layer
;

thus

the

parameters

of

this

second

layer

are

crucial

to

determine

the

response

of

the

agent

to

the

input

taken

by

the

information

layer
.

They

consist

of

four

components
:

Physics,

Cognitive,

Emotional

and

Social

Status
.


3.
Agent

Output

Layer
.

This

layer

is

in

charge

of

modeling

the

set

of

possible

actions

and

the

selection

process,

and

thus

it

produces

the

response

of

the

agent

and

consists

of

two

components
:

The

Behaviour

and

Actor

components
.



14

E
-
Learn 2005


Vancouver, Canada,

October 24
-
28, 2005.



EML

:


Emotional

Markup

Language


Defines

the

emotion

elements

that

affect

the

VH

(virtual

human)

regarding

voice,

face

and

body,

these

elements

are

therefore

inherited

by

the

speech

and

facial

animation

languages
.



The

elements

in

EML

provide

the

VH

with

looks

and

sounds

for

the

specified

emotion
.

The

elements

defined

are

the

following
:

<neutral>

<anger>

<happy>

<sadness>

<fear>

<disgust>

<surprise>


15

E
-
Learn 2005


Vancouver, Canada,

October 24
-
28, 2005.




Attributes

of

EML



Duration
:

Specify

the

time

taken

in

seconds

or

milliseconds

of

the

emotion

existence

in

the

human

being
.



Intensity
:

Specify

the

intensity

of

this

particular

emotion,

either

by

a

descriptive

value

or

by

a

numeric

value
.



Wait
:

Represent

a

pause

in

seconds

or

milliseconds

before

continuing

with

other

elements

of

EML

Temporal

facial

expression

features


Duration

of

Onset
:

how

much

time

is

necessary

for

the

emotion

to

appear?


Duration

of

Apex
:

how

much

time

the

expression

remains

in

this

position
.


Duration

of

Offset
:

how

much

time

so

that

the

expression

will

disappear
.



16

E
-
Learn 2005


Vancouver, Canada,

October 24
-
28, 2005.



We

proposed

a

system

of

collective

and

collaborative

learning

on

the

network

peer

to

peer,

while

being

based

on

PECS

model

integrated

in

the

EECA
.




To

announce

the

difficulties

of

the

learner,




To

make

remotely

the

assistance

more

available

and




To

give

the

more

relevant

assistances

while

resting

on

the

emotive

state

of

the

learner
.



to

return

the

interaction

user
-
computers

more

natural
.




to

react

more

intelligently

and

to

even

anticipate

the

needs

for

the

learner
.


This

system

has

many

advantages

especially

a

better

evaluation,

help

and

support

the

learner

during

the

exercise

in

real

time
.

Conclusion and Future

work


17

E
-
Learn 2005


Vancouver, Canada,

October 24
-
28, 2005.



In

this

paper,

we

pay

particular

attention

to

facial

expressions

of

emotions
.

Facial

expressions

are

characterized

not

only

by

the

signals

of

the

expression

itself

but

also

by

temporal

parameters
:

time

of

appearance

and

disappearance

of

an

expression
.

As

far

as

future

work

is

concerned,

we

plan

to

extensively

test

our

system

on

real

application

cases

in

order

to

validate

the

effectiveness

of

the

approach
.