Designing and Implementing Arabic WordNet Semantic-Based

zurichblueInternet and Web Development

Oct 21, 2013 (3 years and 10 months ago)

150 views

1


Designing and Implementing
Arabic WordNet

Semantic
-
B
ased

Hassanin M. Al
-
Barhamtoshy and Wajdi H. Al
-
Jideebi

Faculty of Computing and Information Technology, King Abdulaziz University
, SA

Abstract

The major aim of this research is to propose
, design

and

implement
linguistic foundations for an
English
-
Arabic dictionary and to explore the demands of this dictionary
,

to be used in the
international languages
,

taken into consideration
both computation and computeriz
ation
. The
major focus of analysis
, design

and implement
ation
, is on the English
-
Arabic pair. However, the
theoretical model within which this analysis has been made,
we

believe, can be extended to other
language pairs.

This proposal presents design
ing

and implement
ing

an Arabic WordNet based on se
mantic. A
relational database is employed to store the lexical and conceptual relations, giving the database
extensibility in Arabic language. The
proposed

model is extended beyond an Arabic replication
of the word/sense relation to include the morphologic
al and lexical roots and patterns of Arabic.
Consequently, t
he model investigates the meaning, synonym, antonym, meronym, hypernym,
hyponym, principle, attribute and pertain
s

structures of the Arabic words.


1. Introduction

Accounts of earlier versions of
the design are given in
[1
-
3], they include:
embedded grammar
tags
,

natural language interaction on the web [1],
facilitating semantic web search with embedded
grammar tags is presented in [2], and
d
ynamic
context generation for natura
l l
anguage
u
nderstanding

is
described as a m
ultifaceted
knowledge a
pproach
.

The EuroWordNet
[4
-
5]
approach to multilingual resource development has emphasized the
separate integrity of the dictionaries in
different languages

and provi
ded an additional bilingual
index to support the search for translations. The effort reported here is on an altogether more
limited scale, and stores the data for different languages in the tables of a single database. Before
we
talk

about languages, let us remember that
the
word is the primary component in any text or
language. The best known way to find out a word is to use a dictionary. We can use it to find out
word

spell, meaning, synonyms
-
prefixes and

suffixes
...etc.


1.1 Introduction to WordNet

WordNet is a lexical database, it provides a large repo
sitory of English lexical items,
which is
available online. The WordNet was designed to e
stablish relations between the main four types
of Parts of Speech (POS): noun, verb, adjective and adverb

[4
-
6]. The synset represents the
smallest unit in WordNet, which describes a specific meaning of a word. It includes the word
itself, explanation and
the synonyms of its meaning.

A specific meaning of one word under one type of POS is called a sense [6]. Each sense of a
word is in a different synset. Synsets are structures containing sets of terms with synonymous
meanings. Each synset has a gloss that
defines the concept it represents. As an example
,

the
words night, nighttime and dark constitute a single synset that has the following gloss [6]: the
time after sunset and before sunrise
,

while it is dark outside. Synsets are connected to one
another thro
ugh explicit semantic relations. Some of these relations (hypernymy, hyponymy for
nouns and hypernymy and troponymy for verbs) constitute kind
-
of and part
-
of (holonymy and
meronymy for nouns) hierarchies [5, 6]. For one word and one type of POS, if there a
re more
than one sense, WordNet organizes them in the order of the most frequently used to the least
frequently used [4
-
6].

2


Synonym
-
substitution algorithms have been developed for the purpose of

matching source
vocabulary terms with existing Unified Medica
l Language System

(UMLS) terms during the
integration process

in [7].

1.2. Introduction to Arabic WordNet

Arabic WordNet is a lexical database, which is structured along the same structures as the Euro
WordNet [4] and Princeton WordNet [
5, 8
]. As mentioned

in researches, WordNet contains
information about nouns, verbs and adverbs in English. Such WordNet is organized in synset
structure. The synset i
s a set of words with the same Part
-
Of
-
S
peech
(POS),

that can be
interchanged in a manner context. As an examp
le, {
ةبرع

,
ةرايس

,
يسكات

,
لقن ةرايس

,

...
etc} form
Arabic synset because they can be used to share the same meaning. Consequently, synsets can be
related to each other by semantic relations, such as synonymy, antonymy, hyponopmy,
meronymy, … etc.
,

as i
llustrated in Figure 1.

















Figure 1: Synsets related to {Car
ةرايس

}


Each of these synsets is related and linked to other synsets as is illustrated for {Car
ةرايس

},
{Vehicle } and { Transport }. Therefore, all word meanings in a language can be interlink
ed
,
interconnected and constitut
e

a relation network (language
-
intern
al relations) or WordNet.


1.3. Arabic WordNet Groups

In Arabic WordNet, we initially worked in five groups: KAU (Saudi Arabia: faculty of
Computing and Information Technology), Cairo University (Faculty of Computer and
Information), Ain Shams University (
Faculty of Engineering), Azhar University (Systems a
nd
Computers Engineering Dept.)

and RDI groups. The Arabic WordNet lexical database will be
integrated with such groups.

We expect that Arabic WordNet will open up a whole range of new tools, applications

and
services in Arabic countries
in

national translation and cultur
al

translation levels. Also, Arabic
WordNet will give non
-
native user
s

the possibility to navigate through the vocabulary of
language
with

new ways. Finally, it will
be
used in information

retrieval, question/ answer
systems, language understanding, expert systems, language modelin
g, document computing

and
{
ةبرع


ةرايس


يسكات


لقن ةرايس

}


ىتوملا لقن ةرايس

}


ةطرش ةرايس

}


ةرايس

Ca爠r

笠噥桩捬h⁽

笠呲a湳灯牴⁽

3


summarizer to automatic translation tools and resources.

In this
paper
, we will use a general
description of the standard database of th
e WordNet and create
a

suitable database of Arabic
WordNet.

The following sections design the proposed Arabic WordNet database (section 2). The general
methodology
,

including database structure of the proposed WordNet is illustrated in section 3.
T
he
WordNet internal relations are described in section 4. The design and implement
ation

of the
proposed model
are
introduce
d

in section 5,

as well as the
overall structure.
Whereas,
section
6

presents conclusion
remarks
.


2
. The Proposed Model Architecture

T
he proposed model is based on morphological Arabic template grammar [13, 14]. Such
grammar can be applied on the word level (morphologic
al

engine).

The morphological engine
employs

the Arabic template grammar and allows affixes processing
to find out diffe
rent alternatives of word input [14].
Lexical disruption engine is used to map
between lexical meanings (words). I
t
is important in case of natural language processing and
machine translation [14].

Also, t
he proposed model is based on interlingua between t
he used distinction languages.
Equivalence relations between the synsets are made explicit in the so called Inter
-
Lingual
-
Index
(ILI) or semantic deep structure.

The model structure consists of languages WordNet
which
use
s

the internal semantic deep
struct
ure (inter lingua). The semantic inter
-
lingua
rep
resents inter
-
relations between languages
WordNet. Figure 2 describes the different modules and their inter
-
lingua relations.

1.

English Word Net
, 2.
Language Translation Module
3.
Arabic Word Net











Figure 2: The Proposed
Model
Architecture

Both the translation module and the semantic relation can be transferred via the equivalence
relations of the ILI
-
records to the language
-
specific meanings, as shown in Figure2. The
semantic relation uses group of language specific lexicons related to the
ILI
-
record concepts.
Such lexicons employ their specific concepts.

Therefore, the main purpose of the semantic relation module is to provide a common framework
for the most impo
rtant concepts between all the WordN
ets. It consist
s

of x basic semantic

Language
Translation
Module






ILI Semantic Relation

Arabic Word Net

1 Language
Dependent

2 Lexical
Descriptions

3. Relation Links

Relation Rules

English Word Net

1 Language
Dependent

2 Lexical
Descriptions

3. Relation Links

Arabic

Dictionary


Dictionary


Lexicon

Arabic

Lexicon

4


disti
nctions that classify a set of ILI
-
records representing the most imp
ortant concepts in the
related W
ord
N
ets. The next sections will describe the semantic relation and its motivation.


2.1 Experimental Relations

The
goal

of the proposed model is to
integrate with the existing WordNet in the other languages;
like English and Euro WordNet. Therefore
,

the methodology is based on using existing
WordNets, and integrat
ing

such WordNets with the Arabic WordNet.

The proposed model experiments this process
on

sequences of string matching and string
manipulation stages
as in F
igure (3).






























Figure 3: Overall Flow for Processing Arabic WordNet relative to other
Languages


The first two stages serve to filter out language terms
that can be found in other W
ord
N
ets. Stage
1 uses conventional matching techniques to find
an
exact string
in other languages (using
languages dictionary).

The second stage uses morphologi
cal rules and Arabic rules to find root(s) and associated
features. However, such stages can only be made by a domain expert, and our concern here is
primarily
on

the results of the synsets stages: stages 3 and 4.


Stage 1
:

Search and/or match for
Arabic word in the standard
translation

dictionary

Stage
2
:

Apply morphological
analysis to find Arabic root
and associated features

Stage
3
:

Search and/or match for
Arabic root in the standard
translation dictionary

Translate the Arabic
word to translated
equivalent word in
other languages

Languages
Dictionary

Stage
4
:

Generate Arabic Synset

Find the equivalent
Synsets for the other
languages

English and
Euro
WordNets

Group of Synsets
in different
languages

Match and link
between equivalent
Synsets

5


2
.
2

Semantic Relation
s

The proposed model

will describe the previous
modules

in more details, but we summarize here
in order to provide clear picture of the events.

The ILI is a list of meaning, taken from WordNet. Each ILI record consists of a synset,
specifying the meaning and the reference to
its source.

Two separate language independent links are linked to ILI records:

1.

Top Ontology, which contains concepts, semantic distinctions
,

such as Object, Substance,
Location, Dynamic and Static.

2.

Domain Ontology, which includes group of meaning
s

in terms

of topics or scripts, such as
Traffic, Road Traffic, Air
-
Traffic, Sports, Hospital and Restaurant.

Any word can belong to multiple synsets, as the following format:

Index Word:



[Word] [POS] : meaning
+

; “example”
+

;

Where:

POS:


[Verb] | [Noun] | [Adjective] | [Adverb]


+: more than one time


WordNet is focused on relationships between synsets, and verbs can be related to verbs, nouns to
nouns, … etc. The semantic relationships available in WordNet are as the following:

1.

All

parts of speech

a.

Synonymy
فدارتلا
: the words that have similar meanings.

b.

Antonymy
داضتلا
/
رفانتلا
: the words that have opposite meanings.

c.

Glossary: is used to store a gloss for every synset.

d.

Similar : connects synsets that have similar meaning.

2.

Verb only

a.

Tr
oponymy
زاجملا
: is a semantic relation of doing something in the manner of something else.

b.

Entailment
اازتتسلاا
: is a relationship between verbs doing something
that
requires doing
something else.

c.

Principle: defines and arranges the relation between verbs
and adjectives.

3.

Nouns only

a.

Hypernymy
امتتشلاا
: refers to a hierarchical relationship between words. (e.g.
furniture

>
chair)

b.

Hyponnymy
نيمضتلا
: is the opposite of hypernamy.

c.

Meronymy
قاقتشلاا
: is part/whole relationship. ( e.g. paper > book)

d.

Attribute:
des
cribes the relation between noun and adjective synsets.

4.

Adjectives only

a.

Participle
: defines and arranges the relation between verbs and adjectives.

b.

Pertain
ق
ّ
زعت
: describes a lexical relation between two words.

c.

Attribute:

describes the relation between noun and adjective synsets.

5.

Adverbs only

a.

Pertain
:
describes a lexical relation between two words.

As stated in MT translation,

the interlingua (ILI) has well known advantages:

1.

New language can be integrated without equivalen
t considerations and relations for other
languages.

6


2.

The ILI can be adapted as a central resource to make matching more efficient and precise.

The equivalence relations of the ILI
-
records can be used by the Top and Domain Ontology to the
language specific m
eanings, as shown in Figure 2. The Top and Domain ontology can be further
inherited by all other related language
-
specific concepts. Consequently, the main purpose of the
Top ontology “is to provide a common framework for the most important concepts in all

WordNets” [
9
].

The Top Ontology consists of “63 base semantic distinctions that classify a set of 1300 ILI
-
records representing the most important concepts in the different WordNets [
9
].


2.
3

Arabic WordNet Synsets

Most synsets are connected to other sy
nsets via a number of semantic relations (based on word
types), and include
[5, 8, 9
]:



Nouns


o

hypernyms
:
Y

is a hypernym of
X

if every
X

is a (kind of)
Y


o

hyponyms
:
Y

is a hyponym of
X

if every
Y

is a (kind of)
X


o

coordinate terms
:
Y

is a coordinate term of
X

if
X

and
Y

share a hypernym

o

holonym
:
Y

is a holonym of
X

if
X

is a part of
Y


o

meronym
:
Y

is a meronym of
X

if
Y

is a part of
X




Verbs


o

hypernym
: the verb
Y

is a hypernym of the verb
X

if the activity
X

is a (kind of)
Y

(
travel

t
o
movement
)

o

troponym
: the verb
Y

is a troponym of the verb
X

if the activity
Y

is doing
X

in
some manner (
lisp

to
talk
)

o

entailment
: the verb
Y

is entailed by
X

if by doing
X

you must be doing
Y

(
sleeping

by
snoring
)

o

coordinate terms
: those verbs sharing a common hypernym



Adjectives


o

related nouns


o

participle of verb




Adverbs


o

Pertain:
root adjectives



The morphology functions of the software distributed with the database try to deduce the
lemma

or
root

form of a
word

from the user's input; only the root form is stored in the database unless it
has irregular inflected forms.


3
. Database Design of Arabic WordNet

The
structure of the Arabic WordNet database is based on the layout structure of the Princeton
WordNet. Therefore, the main semantic relation of the WordNet (Synset) will be taken into our
consideration. However, some specific changes will be made to the
database design, which are
devoted to the following objectives:



to reuse the existing resources of WordNet;



to create multilingual databases [
9
];



to be used in MT (language specific relations in WordNet);



to add
an
equivalent relation for each synset with

respect to Euro WordNet.

7


The database relations can be made across part
-
of
-
speech. WordNet maintains a strict division
between different part
-
of
-
speech
es
.


3.
2.1
Synonym Structure

The first main file is synonym, which stores the synset information of the
WordNet corpus. The
first column
represent
s

SynSetID with 7 digit length, indicating to which synset
the word
belongs
.
As discussed in many researches [
5, 8, 9
, 10
], words belonging to the same synset are
synonyms.
For example
,

synset starting with 122500
0 identif
ies

an Arabic synset (
وتتبقلا
:
acceptance). The words in a synset are numbered serially, starting with one.
Therefore, the
second column represents Arabic synset (may be code or word). The third column is the Arabic
word itself (morphological patte
rn). The fourth column stores the synset category. The synset
categories


in Arabic WorldNet
-

are limited to verbs, nouns, adjectives, and characters.

There
are no any pronouns, prepositions, conjunctions or interjections. Figure
4

shows the proposed
Arab
ic synonym file structure.


SynSetID

Synset

Arabic Word

Category (Type)

1225001

وبقلا

⼠慣ce灴慮ce

باجأ

/ لعف
癥牢

ㄲ㈵〰1

وبقلا

⼠慣ce灴慮ce

رقأ

/ لعف
癥牢


F楧畲u
4
㨠周:⁰牯 潳o搠䅲a扩挠獹湯ny洠晩汥⁳瑲畣瑵牥


I琠浡m 扥 湥e摥搠瑯ta摤d污獴lc潬畭n
,

睨楣栠楮摩ca瑥猠桯眠c潭o潮oa 睯w搠楳i楮i牥污瑩潮l瑯ta 瑥t琠
楮⁡⁴i獴⁣潲灵献pC潮獥煵q湴nyⰠ瑨攠桩t桥爠楳⁴桥畭扥
,

瑨攠浯te⁣潭o潮⁴桥⁷潲搮


W潲摎e琠a汳漠灲潶楤p猠瑨攠
polysemy count

of a word: the number of synsets that contain the
word. If a word part
icipates in several synsets (i.e.
,

has several senses) then
,

typically some
senses are much more common than others. WordNet quantifies this by the
frequency score
: in
which several sample texts have all words semantically tagged with the corresponding syn
set,
and then a count provided indicat
es

how often a word appears in a specific sense.


3
.2.2 Glossary Structure

The second file is used to store a gloss for every synset. Therefore, this file may contain an
explanation, definition and example of Arabic sentences. The structure of this file contains
SynSetID, the second column allocates a gloss in an array structure,

as shown in figure
5
.

Syn
S
etID

Glosses (Gloss [ ] )

1225001

122500
2

1225012


ةوعدلا باجأ


"ةدهاعملا رقأ"

"عقاولا رملأاب مزس"

F楧u牥
5
㨠䅲a扩挠b汯獳ary F楬e⁓瑲畣瑵牥

2.
2.3 Antonym Structure

While semantic relations apply to all members of a synset
,

because they share a meaning but are
all mutually
synonyms
, words can also be connected to other words through lexical relations,
including
antonyms

(opposites of each other

in meaning
) and derivationally related, as well.

Consequently, Antonym file is used to store all relations between words that are antonyms. So
antonym is a lexical relation, it relates two words no two syn
sets. The anto
nym file structure is
shown in F
igure
6
.

8


SynSetID
1

SynSetID
2

1225001

1225112

Figure
6
: Arabic Antonym File Structure

3
.2.4 Hypernym and Hyponym Structure

Hypernym is a relation between synsets, and it is a semantic relation. The hypernym of a
hypernym of a
word

is also a hypernym of the word (hypernym chain). The hypernym file
consists of two columns SynSetID
1

and SynSetID
2
, see F
igure
7
.

SynSetID
1

SynSetI
D
2



Figure
7
: Arabic Hypernym File Structure

3
.2.5 Hyponym Structure

The hyponym is the reversing structure of the hypernym structure (i.e.
,

SynSetID
1

becomes
SynSetID
2

like

F
igure 7.



3
.2.6 Similar Structure

The structure of this file connects synsets

that have similar meaning. The structure consists of
two synsets IDs as arguments, it is like hypernym structure.


3
.2.7 Meronym Structure

Sometimes
a
meronym relation
is
called part/whole relation. This relation

is

only applied in
nouns.
An Arabic word A

is a meronym of another Arabic word B if A is
-
a
-
part of B. The
structure of meronym consists of two synsets IDs as arguments, it is like hypernym structure.
This structure is explained in the original Prolog documentation of WordNet [
8
,10
].


3
.2.8 Attribu
te Structure

Th
is

structure
describes the relation between noun and adjective synsets. An attribute is a noun
that has values and is described by adjectives. For example, the noun (
متجح

size) is an attribute
with the values (
لتيزق

little), (

رييتص

small), (
تض
م

big
(

and (
رتيبك

large).
Therefore, the relation
between nouns and adjectives are defined by attribute operation, and it is similar to antonym,
hypernym, hyponym and similar structures. Consequently, the synset 2225009
-

as an example
-

containing the noun (
نزو

weight) is an attribute relation to the adjectives of the synset

311008,
containing the word (
فيفخ

light). Therefore, the attribute relation between nouns and adjectives is
semantic.


3
.2.9 Principle Structure

This

structure
defines and arranges the
relation between verbs and adjectives. The operation of
this structure arranges the participle relation and therefore describes the verb
-
adjective semantic
relation. The past participle adds “
و
” to the root of the verb, e.g.
,

the verb (
لتعف

did) becomes the

adjective (
وتعف
). The structure of the participle is composed of four arguments to specify two
words
, as shown in the following F
igure 8
.

SynSetID
1

Word
1

SynSetID
2

Word
2





Figure 8:
Principle Structure

3
.2.10 Pertain Structure

This structure describe
s a lexical relation between two words. It indicates where a word pertains
to another word. The structure defines two words by their synset ID and word number.

9


The lexical indicator for th
e first word uses the first two columns;

it can be an adjective or
an
adverb.

If the first word is
an
adjective
-

as a POS, then the second word must be noun or another
adjective. As an example, the first two columns could refer to the adjective lexical word (
اتيموي
),
which pertains to noun (
وتي
), defined by the third and f
ourth columns.

If the first t
w
o

columns
describe an adverb, therefore the assigned second two columns
are

the adjective.

Lexical Word
1

Lexical Word
2

SynSetID
1

Word
1

SynSetID
2

Word
2

3222010

3222011

ايموي

⼠慤橥c瑩癥

ايعوبسأ

⼠慤橥c瑩癥

2111015

2213009

وي

⼠湯畮

عوبسأ
⼠湯畮

4123009

أعفترم
⼠a摪散瑩癥

3123011

عفترم
⼠a摪散瑩癥

Figure 9:
Pertain Structure

3
.3 Database Entity Types

Euro WordNet makes a functional difference between 3 types of entities [
9
,10,11,12
]:

1.

First Order Entity
:
Any concrete entity p
erceivable by senses can be represented in 3
-
D space.

(e.g.
,

object, animal, plant, man, woman, instrument).

2.

Second Order Entity
:
Any static or dynamic situation, which cannot be seen, felt, grasped, hear
d

as an
independent physical thing. They can take place in time rather than exist

(e.g.
,

be, cause, move,
continue, occur, apply).

3.

Third Order Entity
:
Any unobservable proposition independently of time and space, and it can be true
or false or real. It can
be remembered
, forgotten, asserted or denied

(e.g.
,

idea
ةركف
, thought
ريكفت
, theory
ةيرظن
, plan
طيط ت
, information
تاموزعم
, intention
دصقم
, goal
فده
).

4
.

The WordNet Internal Relations

In WordNet and Euro WordNet, the most important relation is synonymy,
which is already
implicit in the notion of a synset. As an example,
F
igure 10

gives the relations encoded in Win
1.5 together with examples for various POS.

Relation

POS

Example

ANTONYM

noun / noun; verb / verb; adjective / adjective

man/woman; enter/exit

HYPONYM

noun/noun

slicer/knife

MERONYM

noun/noun

head/nose

ENTAILMENT

verb/verb

buy/pay

TROPONYM

verb/verb

walk/move

CAUSE

verb/verb

kill/die

DERIVED FROM

adjective/adverb

beautiful/beautifully

ATTRIBUTE

noun/adjective

size/small

SIMILAR TO

adjective/adjective

ponderous/heavy

PARTICIPLE

adjective/verb

elapsed/ elapse

Figure 10: The Relations Encoded in Win 1.5 together with Examples for various POS

The POS contains

:N = noun
,
V = verb
,
Adj

= Adjective
,
Adv= Adverb
, and
PN = pronoun or name
.


5.

Designing
and Implementing
Arabic
-
English WordNet

One way to construct a bilingual WordNet is to have new tables constructed to manipulate
both

WordNet synsets and Arabic words, roots and patterns
,

as well as via synonym, hyponymy and
WordNet relatio
ns. Therefore, the morphological analyzer takes place to analyze each Arabic
word
,

according to Arabic Template Grammar [
13, 14
, 15, 16
-
18
].


10


Figures
1
1

and
1
2

illustrate the overall flow processing of the Arabic word "
ئراق
" and therefore
,

the translated meaning in English
is
"reader".





Figure
1
1
: Overall Flow for Translating Arabic Words "
ئراق
" and "
طط م
" to other Languages



Figure
1
2
: Overall Flow for the Translated Word "reader" to the related synsets


6.
Conclusion

Arabic WordNet is designed to be especially easy to use; it has the simple structure of WordNet
and its underlying representation is based on natural language fragments. Motivated by the range
11


of concepts available in
Arabic dictionaries

commonsense

knowle
dge base, the content of
Arabic
Wor
Net reflects a far

richer set of concepts and semantic relations than those

available in
standard
dictionaries
.

This proposal
has
design
ed

and implement
ed

an
Arabic WordNet based on
multilingual
dictionary and
semantic

relation of standard WordNet
. A relational database
was

employed to
store the lexical and conceptual relations, giving the database extensibility in Arabic language.
The
proposed

model is extended beyond an Arabic replication of the word/sense relation to

include the morphological and lexical roots and patterns of Arabic.

We hope that this paper has encouraged the

reader to consider using
Arabic Word
Net within
their own

projects, and to discover the benefits afforded by such large

scale

semantic resources

like in [19
-
25]
.


Acknowledgements

I would like to thank IT deanship
for

support
,

guidance and insights that
have
contributed
tremendously to this paper. Also
we

would like to acknowledge
Basil A. Ba
-
Aziz and
Seef Al
-
Harthi for
their
programming
effort
s
.
Many thanks to: WordNet Princeton
,
for using some results of
their research papers.

This work has been supported, in part, by KAU Award # MT
-

01
-
01
-
14230.


References

1.

G. Dorai, and Y. Yacoob, 2002. "
Embedded grammar tags: advancing natural language
interaction on
the Web
", Intelligent Systems, IEEE Volume 17,


Issue 1,


Jan/Feb 2002

2.

G. Dorai, and Y. Yacoob, 2001. "Facilitating Semantic Web Search with Embedded Grammar Tags
"
Workshop on

E
-
Business & the Intelligent Web
,
USA, August 2001
.

3.

Samuel W. K
. Chan, 2003. "Dynamic Context Generation for Natural Language Understanding: A
Multifaceted Knowledge Approach",
IEEE Trans. Syst., Man, Cybern.
, vol.33, No. 1 Jan. 2003

4.

J. Gonzalo, 2004."Sense Proximity versus Sense Relations", The Second Global Wordnet

Conference

5.

Miller G. A., Beckwidth, R., Fellbaum, C., Gross, D. and Miller, K. J.,1999. “Introduction to
WordNet: An On
-
line Lexical Database”,
International Journal of Lexicography
, Vol 3, No.4 (Winter
1990), pp. 235
-
244.

6.

Thanh Dao, An improvement on ca
pturing similarity between strings,
An improvement on capturing
similarity between strings, codeproject.com, 2007.

7.

Kuo
-
Chuan Huang , James Geller, Michael Halper, Yehoshua Perl, Junchuan Xu, (2009). Using
WordNet synonym substitution to enhance

UMLS source

integration,
Artificial Intelligence in
Medicine (2009) 46, 97
-
109
.


8.

Fellbaum (2004),

WordNet 2.0. Available from:
http://wordnet.princeton.edu/oldversions


(
accessed
14 August 2009)
.

9.

Euro WordNet

Web Site:
http://www.illc.uva.nl/EuroWordNet/
,
(
accessed 1 August 2009)

10.

Sarah Witzig, Accessing WordNet from Prolog, Artificial Intellegence Center, the University of
Georgia, 2003 May 8.

11.

S. Chua, and N.

Kulathuramaiyer, 2004. "Semantic Feature Selection Using WordNet ", Proceedings
of the IEEE/WIC/ACM International Conference on Web Intelligence (WI’04) 0
-
7695
-
2100
-
2/04

12.

William J. Black and Sabri El
-
Kateb, A Prototype English
-
Arabic Dictionary Based on W
ordNet,
UMIST, Department of Computation, Manchester, M60 1QD, UK, 2005.

13.

Hassanin M. Al
-
Barhamtoshy, Khaled O. Thabet and Basel Ba
-
Aziz, ‘Arabic Morphology Template
Grammar
-
Based”, the 7
th

Conference on Language Engineering 2007, Cairo.

14.

Hassanin M.
Al
-
Barhamtoshy, Mostafa S. Saleh and Reddah A. Al
-
Kheribi, "Infra Structure for
Machine Translation”, the 7
th

Conference on Language Engineering 2007, Cairo.

15.

H. Yang and M. Zhang, 2005. "Ontology
-
based Resource Descriptions for Distributed Information
12


Sour
ces", ICITA'05.

16.

Black, W. J., El
-
Kateb, S.: Towards the design of an English
-
Arabic termonological and lexical
knowledge base. In Proceedings of the Workshop on Arabic Natural Language Processing, ACL
-
2001,
Toulouse, France (2001).

17.

Denness, S. M.: A Design

of a Structure for a Multilingual Conceptual Dictionary. MSc dissertation,
UMIST, Manchester, UK (1996).

18.

Ramsay, A. and H. Mansur.: Arabic Morphology: a categorial approach. Proceedings of the
ACL2001 Workshop on Arabic Language Processing: Status and Pro
spects, Toulouse, July 6th, 2001.

19.

Sierra, G. and McNaught, J.: Design of an onomasiological search system: A concept
-
oriented tool
for terminology. Terminology 6(1), 1

34. (2000).

20.

H Liu and P Singh
"
ConceptNet
-

a practical commonsense reasoning tool
-
kit” B
T Technology
Journal, Vol 22 No 4 (October 2004).

21.

Ellen Riloff and Jessica Shepherd “A Corpus
-
Based Approach for Building Semantic Lexicons”
Department of Computer Science, University of Utah.

22.

Lim Beng Tat, Dr Tang Enya Kong and Dr Guo Cheng Ming “Building

a Semantic
-
Primitive
-
Based
Lexical Consultation System” Unit Terjemahan Melalui Komputer,Universiti Sains Malaysia,11800,
Pulau Pinang.

23.

Branimir K. Boguraev“Building a Lexicon: The Contribution of Computers” Lexical Systems Project,
IBM Research, Thomas J
. Watson Research Center Yorktown Heights, New York 10598, USA.

24.

Dustin Smith, Advisor: Stan Thomas, Ph.D “Investigating ConceptNet”Wake Forest Department of
Computer Science December 2004.

25.

Satoru Ikebana “Toward the Realization of Typological Semantic
Pattern Dictionaries for MT”

Faculty of Engineering, Tottori University, Japan.