An Intelligent Problem Solving Environment in the Domain of Electrical Engineering

(1)

Ätifcial lntelbsence in hl,catibr

HU

HoWe

dol

pns

₎

IOSPrc$,20A3

An Intelligent

Problem Solving Environment

in

the

Domain

of

Electrical Engineering

Vera

YAKIMCHUK,

Hilke GARBE, HeiD-Jürsen THOLE

hntotiun

OFFIS e.y.,

hche

eg2,26121Oldenbury, GERMANY

{yakinchuk, Gdbe, Thore}@ol1is.de

Claus

MöBUS

Ca

vot

Ossiealcy

U

tueßity,

Depd

hekt

ofconpatihg

Sck

ce, 26 t I

t

O ldenbwg, G E RMANY

C I a u s. Mo e b

ß

_@ikfo

m

a t i k un i - o I de n b

ft

g. de

Edwin WACNER

Technical

U

tuersitr

oflneMu,

Ißtitute

ofcenenl

Ebctrical Engineerins,

PF 1005

65,98681Il

e

au, GERMANY E dw

iL

Wagw r@.tü-il

ne

at.de

Ahstract This pap€r presenls 6i1eET

d

iftellisent problen $lving

€.vnomül

(IPSE) lor a lopü aEa

ton

ihe bßics ol gmenl el@üical enginee.i.s. Mil€ET

,Lppon rre

edq

Ltuoughoi üe *\ole

plM$

oi p.oblem .olv 1C

Or

qorl

ß

baed on .oeniti\e nience ?,e/.1. rlsP

Dt

lheoD

_{l.l)olougrcLp}

donl.

d€v.lopmor of IPSES in vnios domains [1,2]

a

well

6

m üe conpetence of rh€ partn€r miversiti* in

gdeal

elecriel ensin€eri.s d on dret etaboiared rak

collectons. As no fomal speificatio. of lhe 14ks exis6,

w

classified tlE task

according ro rheir infomal spsified

_süls

and elaboräted a new nodet (catled Y,

Model)

dsribins

the proces of domain speciäc

lak

solving. This mod.l is implemenred in Prolog as a set of soal,m€d-rclatioß

(cMrr)

_[3],

dd

auos

e

elabontion of lbe plming objetives

üd

synbolic conpütation.

IlEs

cMtu

l:m

lhe core ol ou _smmtive€xpen sygen (CXIS) lnat prcvides th€ adaptive support

üüil€ sotving uiks,

Kqrords.

Inrelligol probleft elvins mvnomenrs, adaptive ht?oüeses t$trng, eledrical ensin€eri.g education, nodels of r$ks, tnowledse represmtatio4 cog.ilive diagnsis, adaptive envnomenls

Most

of

the €leaming

developments

in

the

domain

of

electrical engineering have an

instructional chamcter- Even

ifthey

mntain tasks the leamer has to solve they don'r analja€

steps

ofthe

solution path or incomplete solution proposals. So situarion,adapted help can't be

given

ifthe

user is in an impasse while solving a task. Other prograns used by studenls, like

compuleralgebrasystems(CAS)mdprofessionalcircuiteditors,c'tanalyseifequationsor

circuits belong

to

the solution

for ä

given lask.

As

both circuils and equations build the

solution

of

mosl

of

ou

tasks

an

environment

with

editors

for

circuits and formulas is

(2)

519

IPSES I I

]

belons to the class of constructivistic le3ming environnents. They enabl€ students to solve domain relevant tasks. During problem solving the leamer can denand adaplive help creat€d by a GXPS. This GXPS is abi€ to analys€ the solution proposals and

il

is lherefore

abl€ to r€spond adaptively to student bwotheses.

The mileET tasks are taken from lask colleclions of the partner

$iveßities

d

will

be saved

in the _{Taskweb-Database [4].}

1. Knowledge Representation

rnd

Hypothses Testing

After

d€tailed

anal$is

of

the task collections the

Y-Model

for

the representation

of

the

process ofdomain specific task solving was elabonted. Any supported task car be understood

as relation over the sets of circuits (C.), parameteß (P), mathernatical descriptions

(M),

and domain conceprs

(CD.

'rhe circuit

lopolos/

information is srored in the

C'sel

s)mbolic änd nümeric parameter in the P-set. and malhematical equations in th€ M-set- According to a task

goal several

Y-Tpes

of

our tasks can be

built.

Malhematical equations, parameters and

circuits

parts of all fiese sets may be ä task's goals. These sets P, C, and

M cm

be derived

fiom each other by application

ofCF.

For example, by application of th€

Kirchhofs

laws on

a

given network tbe

KCL

equations can be obtained.

This

d€rivalion represents

üe

task solving proc€ss

of

an expert and can be produced by our IPSE

by

means

of

goals-means-relations (GMR) .

Owing to GMRS and to a sp€cial GMR-Mela-Iilerpreter the syst€m

will

be able to examine hwotheses ofthe students, and to possibly complete solution proposals, 1oo. GMRS assign lhe means (solutions) to th€ goals

(task

and subtaskt. In our IPSE

all

means ar€ parts of the three sets (C), (P) and (M)

ofthe

former described Y-model. As our OMR-rules are very fine-elaioed th€y caft be combined to get a very

ldge

set

of

different solutions, containing also

2.

Working Eovironnent rnd

Hypothes€s Testing

The working environment (figures

l,

2) consists

of

several scalable areas, here shown in a

[rF*kbwiq,ryre5:b. FnH

DiEod'@tuc dri!6d,

(3)

In the task area

(figure

l,

left) the task settings and the

infomätion

about circuit elemenls are shown. In the middle of lhe window the circuit editor is located, where ihe shrdent can

work on the task's

circuit

(set or change the properli€s

of

cir€uit elements,

rransfom

or simplify the circuir). On the

right

side there is the worksheet area.

It

cootains the solulion

proposal

of

the student.

lf

the ,eamer thinks the circuit he has ediled belongs to the task

solution, he can copy

il

into

the lvorksheet.

He

also can

write

formulas

wirh a

special

implenented

fomula

editor. The

system's feedback can

be

seen

at the

bottom

of

rhe window or in amessage window.

The leamer may ask tbe

slst€n

for

an evaluation

by

narkins

his

proposal even

if

ir

is incomplete.

To

prove the conectness

of

the proposal the s)6tem tries

to

embed

it

in

a

dynamically generated solution- In case

ofan

inconecr solution proposal the system gives error feedback- Now the user has to

fomulate

a new hlpothesis by marking the parts of his proposal which he assunes to be conect. The system gives positive feedback

only

if

the

hpoth€sis

cao be ernbedded

in

a correct solulion.

h

this

case the solution trace

ofrhe

program can be requested by the student (fi8ure 2)

for

getting help how to €omplere his proposal.

If

the studenl läcks declarative knowledse, he can stan the web,based leaming modules

of

lhe

partner

univeßities. Tbe

appropriate

linls will

be

offered

in

rhe help

3. Conclusiotr

Oür learning environrnent is meant for beginners who do not have enough experience with the professionäl software packages such as CircuitMaker or Mathematica, and who need support

by solving tasks. Th€ problem solving environment

is

a&pied

to

the

rask

of

our partner universities. It can be used to solve differenl other tasks, too, as long as these tasks lie within üe predefined topic fields. At lhe prcsent tasks e.g. to such _{topics like basic el€ctrical circuits,}

Kirchhors

laws, equivalent circuits are süpported. Further on, the working environnent c

be

used task independ€nlly.

The

student

may

use

the

system

äs än

experjmedal work

env;ronment änd explore any circuit fiom _{the knowledge domain implemented in the slstem.}

Our

leaming environn€nt can be used both

in

direcl study and

for

self leaming. Slstem explanälions

and

comnent€d solutions are

helptul during

preparation

for

€xarninarion. Comprehensive lesting and evaluation of our IPSE

will

take place ar our parrner

uiversities

ailer completion ofthe first protog?e.

tll

MÖBUS. C. d al., Towards m Al-Sp{iricarion ottntetligenr Dislributed Leming Ehvtonnmrs, AI Joumal I$ue l/03 "howledge Modelins

ed

Knovledge Comunicatio. in

L@ins

Scendios , rSSN 0933 1875, 2003

t2l MÖBUS, C., SCHRöDER, O. & THOLE, H.,J., Diagnosihg

dd

Evalualing lho Acquisition prccBs of Progrmine Sch€mar4 in LE. Creer, C. McCaUa (eds), Studenl Modelling: The Key to rndilidutized kowledge

Bed

lßlruction, Berlin: Sprinser CNATO ASI Series F: Conpurer and Systems Sciencs, Vot.

125). pages 2ll-264, 1994

[]l

LÜDTKI, A., MÖBUS, C., THOLE. H.-J., Copilile Modellins Appmach to Diägnose Ovq-sinptificarim

i.

Sinülatio.-Bed Trainins,

i.

Sl.A. Cmi, C. courderes

&

F. p@sucu (eds), rntelligmi Tuloins

S'$ems! Proceedinss of the 6rh Intemational Corfmncq ITS2002. Bimilz, Fdnce Md

Se

Sebasliaa Stain, B€rlin: Spnngd, Lectue Notes in Cohplter Sciencq LNCS 2163, ISBN 3-540-43750 9, psses 496

-506.2002

(4)

Artificial

Intelligence

in

Education

Shaping the

Fuhre

of Learning through

Intelligent

Technologies

Edited

by

Ulrich

Hoppe

UniveßftAt Duisbutg-Esserr, Genfta

y

Felisa

Verdejo

Uniyersidad

Nacional

de

EdlAaciöh

a

Distancia, Madrid, Spain

and

Judy

Kay

Univers

it,

of

Sydnqt,

Alstralia

tos

P.e!s

+==

clmha

(5)

O 2OOl.

fte

aulhqi mfltion d in rh. tabk of@nte.ts

AU rig!ß

resftd.

No p.n of üi, 6mk

try

b. Ep@dn@4 sio@d ir .

Eri*l

s)6rdt! or ba]eided,

inany{{morby

y

mds

wirholt

_lfid

qits

Ddnission fioh dr pühlislrr. ISBN I 58603 356 5 (lOS

PGt

ISBN 4 274 90600 0 C3055 (Ohuh.)

Libnry

ofc6gt6

Conlrol Nurnd: 200310603?

fd+31206203419

Ditnibstor

k

the UK

a

l

lrels"d

Ditttibrtat in tte USI and Cnnad!

IOS Pr.ss4jvis

Ma*etins

IOS

16,

Irc.

73 Lim

wdt

5?95-GBukcco iePdkMy

Heddington

BüLq vA 22015

ONlodo)C

?AD

USA

Engl&d

fd; +l ?03123 3663

f!*

+44 1865 ?5

00?9

eMil

i6bdl6@iosp6s$e

Dßnibutü in Cemdn , Aßria atul

SwiEqlüd

DisEibttot

t

J4Pan

D44103 Leipzis Chjlod!-k\ Toklb 101-8460

LECAI NOTICE

Thc publ6h.r is ml Bpotuible for ü.

w

wnich miSh be rsd. ofile fol@ing infomliotr PRINIED IN TIIE NETTIERI-ANDS