Semantic Web Approach in Designing a Collaborative E-Item Bank System

steelsquareInternet και Εφαρμογές Web

20 Οκτ 2013 (πριν από 3 χρόνια και 8 μήνες)

250 εμφανίσεις

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Semantic Web Approach in Designing a
Collaborative E
-
Item Bank System

Heung
-
Nam Kim

With
Ae
-
Ttie Ji
, Soon
-
Geun Lee, and
Geun
-
Sik Jo

Dept. of Computer Science & Information Engineering

Inha University, Korea

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Outline


Introduction


Background and Related Works


E
-
Item Bank System based on Semantic Web


Experimental Evaluation


Conclusion & Future Work

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Introduction


Educational environment and the fundamental
paradigm have been shifted


E
-
Learning, Web
-
based online
-
assessment …



Item banks

is repositories of learning objects,
particularly
assessment questions


Various factors are now coming together!!


Organizing, Maintenance, Availability …


Searching, Interoperability, Reusability …


Quality assurance, Copyright …

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Introduction
(Cont’)


Goal of Designing Item banks


Organizing item banks with
well
-
defined semantics


Clear definition of
relationship

between items and
users (teachers or students) or between teachers
and students


Accurate
searching

not only item itself but also
extra
-
information related to item and user


Promoting item
reusability

and
sharing

Semantic Web technologies

(Ontology, Domain rules, Knowledge inference …)

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Background Research


Item Banks (Item Pool)


Collections of items, often produced collaboratively
across a subject domain that can be grouped
according to difficulty, the type of skill or topic


Item Bank

Item Pool

Item

Item

Item

Assessment
-
specific

Subject
-
specific

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Background Research


Semantic Web


Extension of the current web in which information is
given well
-
defined meaning, better enabling
computers and people to work in cooperation (
Tim
Berners
-
Lee, 2001)


Knowledge representation in SW


Ontology
-

OWL
, RDF


Rules


RuleML,
SWRL
, ORL


Reasoning or Inferencing in SW


Jena, OWLJessKB, JESS
, F
-
OWL

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

E
-
Item Bank System based on Semantic Web


Architecture of E
-
Item Bank System based on Semantic Web

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Ontology Modeling for E
-
Item Bank


Concepts (Classes)


Sets of individuals with common characteristics


Item, Topic, Profile, Curriculum, Answer



Individuals


Represent objects in the domain, Specific things



Properties


Object properties : Link two individuals together


Data properties : Link individuals to primitive values

(integers, floats, strings, booleans etc)


Name, affiliation, submitDate, stem, ,GUID…

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Ontology Modeling for E
-
Item Bank


Ontological structure in E
-
item Bank System

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Ontology Modeling
(Cont’)


Object Properties Characiteristics

Object Property

Domain

Range

Characteristics

Inverse

hasCurriculum

Teacher

Curriculum

isCurriculumOf

hasFriend

Student

Student

Symmetric,

Transitive

hasInterest

Student

Topic

hasProfile

Item

Teacher

Functional

isProfileOf

hasSameInerest

Student

Student

Symmetric,

Transitive

hasSameInerest

Teacher

Teacher

Symmetric,

Transitive

hasStudent

Teacher

Student

hasTeacher

Student

Teacher

hasTopic

Item

Topic

isTopicOf

hasTopicBase

Curriculum

Topic

isTopicBasdOf

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Ontological Relationship for
Cooperative Education


Relationship between items and users
(teachers and students) or between users


Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Domain rules for Cooperative Education


Rule
-
1: hasTeacher(?x, ?y)


hasTeacher(?z, ?y) →hasFriend(?x, ?z)

Rule
-
2: hasTeacher(?x, ?y)


hasFriend(?x, ?z) → hasTeacher(?z, ?y)

Rule
-
3: hasInterest(?x, ?y)


hasInterest(?z, ?y) → hasSameInterest(?x, ?z)

Rule
-
4: ……



Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Semantic Searching


OWLJessKB


Memory
-
based reasoning tool for description logic, OWL


Uses the Java Expert System Shell (JESS) as underlying
reasoner








The semantic element continuously goes through a reasoning
process through rules with the facts in JESS.


Semantic Searching based on JESS Inference

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Semantic Searching
(Cont’)


OWL rules


provides the meaning of ontology composed of OWL
to JESS as a rule.

(defrule OWL_inverseOf


(triple (predicate "http://www.w3.org/2002/07/owl#inverseOf")


(subject ?x) (object ?y))


(triple (predicate ?x) (subject ?u) (object ?v))


=>


(assert (triple (predicate ?y) (subject ?v) (object ?u)))

)


The property characteristic
inverseOf

for JESS triple

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Semantic Searching
(Cont’)


Domain rules


Representing SWRL (Semantic Web Rule Language)


Inferring new knowledge from existing OWL item banks

No

Example of Domain rules

Rule
-
1

hasTeacher(?x, ?y)


hasTeacher(?z, ?y) →hasFriend(?x, ?z)

“If Y is a teacher of student X and student Z, Then X and Z is a classmate.”

Rule
-
2

hasInterest(?x, ?y)


hasInterest(?z, ?y) → hasSameInterest(?x, ?z)


If a student x, has interest on y, and a student z, also has interest on y as well, x
and z have common interest of both
.”

Rule
-
3

hasTopic(?x, amylase)





hasTopic(?x, catalyst)

enzyme(?x)


isTopicOf(?x, ?y)





isTopicOf(catalyst, ?y)

“If a question is related to amylase topic, Then it is also related to catalyst topic”

Rule
-
4

hasTeacher(?x, ?y)


hasFriend(?x, ?z) → hasTeacher(?z, ?y)

“If Y is a teacher of student X and student X is classmate with student Z,

Then Y is also a teacher of student Z.”

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Semantic Searching
(Cont’)


Convert domain rules to the triple form of JESS

(defrule rule
-
2


(triple


(predicate "http://eslab.inha.ac.kr/ITEMPOOL.owl#hasInterest")


(subject


?z)


(object


?y)

)



(triple


(predicate "http://eslab.inha.ac.kr/ITEMPOOL.owl#hasInterest")


(subject


?x)


(object


?y)

)


=>



(assert





(triple


(predicate "http://eslab.inha.ac.kr/ITEMPOOL.owl#hasSameInterest")


(subject


?x)


(object


?z)

) )

)

The triple type of JESS for Rule
-
2

hasInterest(?x, ?y)


hasInterest(?z, ?y)



→ hasSameInterest(?x, ?z)

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Semantic Searching
(Cont’)


Convert OWL fact to the triple form of JESS


(assert



(triple



(predicate

"http
:
//www
.
w
3
.
org/
2000
/
01
/rdf
-
schema#subClassOf")



(subject


"http
:
//eslab
.
inha
.
ac
.
kr/ITEMPOOL
.
owl#Selection")



(object


"http
:
//eslab
.
inha
.
ac
.
kr/ITEMPOOL
.
owl#Item")



)

)

(assert



(triple



(predicate

"http
:
//www
.
w
3
.
org/
1999
/
02
/
22
-
rdf
-
syntax
-
ns#type")



(subject


"http
:
//eslab
.
inha
.
ac
.
kr/ITEMPOOL
.
owl#Item")



(object


"http
:
//www
.
w
3
.
org/
2002
/
07
/owl#Class")



)

)


Selection

Item

subClassOf

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Experimental Evaluation


Modeling ontology and defining rule using Protégé 3.1


10 teachers and 35 high school students participated in
the experiments


Each teacher formulated 50 questions (total. 500)


Obtained basic information from them (profile, interests …)



50 queries with simple type or complex type were
generated and tested

No

Query examples

1

Search items submitted by a teacher whose curriculum is biology

2

Find teachers who have a interesting topic corresponding with In
-
Kyung Bae

3

Find teachers teaching students interested in allrosteric


4

What is the student’s name who is classmates with O.J. Oh and has the same interests?


5

Fine Items that include the word ‘photosynthesis’ and are formulated by a teacher who has
an interesting topic corresponding with Heung
-
Nam Kim

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Experimental Evaluation


Comparison of three type of E
-
item bank system


RDBIB
: RDB
-
based E
-
item bank


SWEIB
: Semantic web
-
based E
-
item bank without domain
rules


SWEIB+SWRL
: Semantic web
-
based E
-
item bank with domain
rules



Evaluation Metrics

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Experimental Results


Comparison of precision and recall of RDBIB,
SWEIB, and SWEIB+SWRL

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Analysis of experimental Results


Synonymic or ambiguous vocabulary



In SWEIB, SWEIB+SWRL, item related to ‘
photosynthesis’

was
also searched. (sameAs)


Reasoning new facts by ontology


Mr. Kim

is classmate with
Miss. Ji


Miss. Ji

is classmate with
Mr. Lee


RDBIB did not find
Mr.Lee
, SWEIB & SWEIB+SWRL did
(Transitivity property)


Inference new facts by ontology & domain rules


Mr. Kim and Mr. Lee is interested in ‘nephron’


SWEIB+SWRL found the result ‘
Mr. Lee
’,
but

the others didn’t


Inference new fact: Mr. Kim and Mr. Lee have the common
interest (hasSameInterest)

Find items related to ‘
assimilation’

Search the classmates of Mr. Kim

Find the classmates who has common interest with a student Mr. Kim

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

Conclusion


Designing item banks with semantic web
technologies


induce users to cooperative education


provide semantic searching not only item itself but
also extra
-
information related to item and user


promote item reusability and sharing



Future study: Semantic Web Service


Integration of distributed item banks


Automated item selection for a personalized
assessment

Intelligent E
-
Commerce Systems Lab

INHA University

33rd International Conference on

Current Trends in Theory and Practice of Computer Science

nami@eslab.inha.ac.kr


http://eslab.inha.ac.kr

Thank you !!!