Volume 05 Issue 12-2024
49
CURRENT RESEARCH JOURNAL OF PHILOLOGICAL SCIENCES
(ISSN
–
2767-3758)
VOLUME
05
ISSUE
12
Pages:
49-54
SJIF
I
MPACT
FACTOR
(2022:
6.
041
)
(2023:
7.
491
)
(2024:
8.235
)
OCLC
–
1242423883
Publisher:
Master Journals
ABSTRACT
This article analyzes the concept of a thesaurus and its capabilities, comparing the differences between the notions
of WordNet and a thesaurus. It highlights the structure, composition, and challenges faced in creating linguistic
resources like the Turkish WordNet, Arabic WordNet, and the Uzbek UzWordNet, which is based on WordNet.
Preliminary linguistic models serve as a foundation for the Uzbek WordNet. Additionally, ideas regarding linguistic
research aimed at creating the Uzbek UzNet network are presented.
KEYWORDS
Thesaurus, WordNet structure, WordNet, KeNet, AWN (Arabic WordNet), merging, splitting.
INTRODUCTION
One of the main issues in global linguistics in recent
years has been the rapid development of universal
language resources. These include various resources
such as software and linguistic databases (lexicons,
lexical data repositories, grammars, and corpora
marked in different ways), which are used in both
research and industrial applications. At present,
comprehensive
ontologies
are
available
for
knowledge-based NLP tasks. Princeton WordNet,
BabelNet, FramNet, and the European WordNet are
among the most renowned ontologies.
Although manually constructing WordNet for any
language is a time-consuming and labor-intensive
process, it is notable for its accuracy. Existing
dictionaries in a language are limited to augmenting
databases. For other languages, some authors have
emphasized the importance of using XML-type
specially extended languages, which, in turn, facilitate
Research Article
WORDNET
–
A LEXICAL DATABASE FOR LINGUISTIC ONTOLOGIES
Submission Date:
December 02, 2024,
Accepted Date:
December 07, 2024,
Published Date:
December 12, 2024
Crossref doi:
https://doi.org/10.37547/philological-crjps-05-12-11
Pulatova Gulhayo
Independent researcher at Namangan State University, Uzbekistan
Journal
Website:
https://masterjournals.
com/index.php/crjps
Copyright:
Original
content from this work
may be used under the
terms of the creative
commons
attributes
4.0 licence.
Volume 05 Issue 12-2024
50
CURRENT RESEARCH JOURNAL OF PHILOLOGICAL SCIENCES
(ISSN
–
2767-3758)
VOLUME
05
ISSUE
12
Pages:
49-54
SJIF
I
MPACT
FACTOR
(2022:
6.
041
)
(2023:
7.
491
)
(2024:
8.235
)
OCLC
–
1242423883
Publisher:
Master Journals
working with structured data [1]. Princeton’s model
has been utilized in creating WordNets for Turkish
(Turkish WordNet), Arabic (Arabic WordNet), and
many other languages. KeNet, another WordNet
created for the Turkish language, is based on the
Turkish lexicon.
METHODOLOGY
In 1985, a group of linguists and psychologists at
Princeton University conducted research for the
WordNet lexical database project, also referred to as
"lexical ontology." The project was inspired by
psycholinguist George Miller's experiments with
artificial intelligence aimed at understanding human
semantic memory. According to Miller, WordNet
contains approximately 95,600 word forms (51,500
words and 44,100 multi-word expressions) with 70,100
meanings represented through synonyms. Among
these, approximately 57,000 are nouns, of which
47,000 have recorded meanings.
The variability in the number of entries is due to the
dynamic nature of the online system, where these
figures are constantly evolving. Unlike traditional
dictionaries, WordNet is exclusively composed of
nouns, verbs, adjectives, and adverbs [2]. WordNet is a
psycholinguistic lexical database created at Princeton
University, widely used in language engineering
research [3].
WordNet serves as a lexical database for the English
language, where nouns, verbs, adjectives, and adverbs
are grouped into synonym sets (Synsets) representing
a specific concept. Synsets are interconnected based
on conceptual-semantic and lexical relationships.
WordNet is one of the most user-friendly tools for
natural language processing and computational
linguistics. While it resembles a thesaurus in some
aspects, significant differences exist. Notably, a
thesaurus does not include semantically related words.
In contrast, WordNet groups words based on semantic
relationships such as synonyms, hyponyms, and
meronyms. The central relationship in WordNet is
synonymy. Synonyms are organized into Synsets,
which include brief definitions and usage examples.
Therefore, WordNet can be seen as a combination or
extension of a dictionary and thesaurus.
Analyzing the WordNet system using the noun
category as an example reveals that the database
contains
over
80,000
nouns.
This
includes
opportunities to use lexemes as speech units or word
combinations. WordNet is among the most innovative
achievements in computational lexicography [3].
DISCUSSION
Robert M. Loseelar’s research on thesaurus
development discusses methods for incorporating
thesaurus features into indexing and retrieval systems.
He emphasizes that understanding how people use
terms simplifies this problem and facilitates decision-
making, which should rely on hypotheses about future
research by the author or others [4].
Traditionally, dictionaries store information about
words and their definitions. With the development of
natural language processing research, the need for
machine-readable dictionaries has emerged [Miller,
1995:39-41]. To meet this need, lexical networks have
been created, adapting lexical data for modern
computing. WordNet is one of the earliest, most
comprehensive lexical databases for the English
language, serving as a prototype thesaurus in some
sources and as open electronic ontology in others.
WordNet is part of the lexicographical class of
ontologies and is extensively used in information
retrieval. Thousands of experiments in this domain
have been conducted using its framework [5].
WordNet, in a broader sense, is a comprehensive
dictionary built around definitions of specific word
meanings.
Volume 05 Issue 12-2024
51
CURRENT RESEARCH JOURNAL OF PHILOLOGICAL SCIENCES
(ISSN
–
2767-3758)
VOLUME
05
ISSUE
12
Pages:
49-54
SJIF
I
MPACT
FACTOR
(2022:
6.
041
)
(2023:
7.
491
)
(2024:
8.235
)
OCLC
–
1242423883
Publisher:
Master Journals
WordNet’s foundational concept is the Synset—
a
group of synonymous words sharing the same
meaning. Given that words can have multiple
meanings, a lemma may belong to several Synsets
representing its various senses. Additionally, WordNet
defines relationships among Synsets, such as
hyperonymy, hyponymy, and meronymy. Synsets form
the core of WordNet and have inspired the creation of
similar electronic resources for dozens of other
languages worldwide.
For instance, Turkish WordNet (KeNet), Finnish
WordNet (FinnWordNet), Polish WordNet, Norwegian
WordNet, Danish WordNet, and French WordNet
(WOLF) have been developed. Below, we will examine
the development methods and structure of Turkish
WordNet (KeNet), Arabic WordNet (AWN), and Uzbek
UzWordNet, which were modeled on English WordNet
resources.
Currently, two word networks exist for Turkish:
BalkaNet TR-wordnet and KeNet. Unlike BalkaNet,
which was based on the Princeton WordNet model,
KeNet was constructed from scratch using a bottom-
up approach and is the most comprehensive word
network for Turkish, containing 76,757 Synsets. The
BalkaNet Turkish WordNet (TR-wordnet) was the first
WordNet created for Turkish and encompasses
approximately 14,626 Synsets and 19,834 internal
semantic relationships [8].
In the creation of KeNet, managing the semantic
relationships within Synsets posed significant
challenges. To address these, two additional principles
were implemented: merging and splitting Synsets.
During the merging process, multiple Synsets were
combined into groups, resulting in the addition of new
definitions for 10,612 Synsets.
As with other WordNets, the first step in developing
KeNet was creating Synsets, defined as groups of
interchangeable words sharing the same meaning and
part of speech (POS). A sample structure of a Synset in
KeNet is as follows:
Table 1. Sample Synset.
Using Examples from the Table to Illustrate KeNet’s
Frequent Speech Parts
KeNet includes four major parts of speech frequently
encountered: noun, adjective, verb, and adverb. For
each example in the table, the first column represents
the Sinset ID. The identifiers are separated by a "-"
symbol, indicating the Sinset POS (n
–
noun, v
–
verb, a
–
Volume 05 Issue 12-2024
52
CURRENT RESEARCH JOURNAL OF PHILOLOGICAL SCIENCES
(ISSN
–
2767-3758)
VOLUME
05
ISSUE
12
Pages:
49-54
SJIF
I
MPACT
FACTOR
(2022:
6.
041
)
(2023:
7.
491
)
(2024:
8.235
)
OCLC
–
1242423883
Publisher:
Master Journals
adjective, adv
–
adverb). The second column lists
synonyms within the respective Sinset.
According to KeNet’s creators, some necessary
components were missing in certain Sinset groups
within WordNet. While KeNet addressed some of these
deficiencies, others remain unaddressed. Existing
Sinset group members in KeNet were merged with
WordNet synonym sets. Sinsets absent in KeNet were
added to existing Sinsets as separate components. This
process enriched 122 Sinsets with new members [5].
Splitting Process in KeNet
During the splitting process, Sinsets with different
meanings were separated, and new Sinsets were
created for each group. Semantically unrelated or
incorrect Sinsets were also separated, and new
definitions were assigned to them. These Sinsets are
provided in Google Sheets tables for clarity.
Table 2. Number of Sinsets in KeNet
Specifically, KeNet has served as a foundational lexicon
for projects like the Turkish PropBank TRopBank (Kara
et al., 2020), Turkish SentiNet HisNet (Ozcelik et al.,
2020), Turkish FrameNet (Marsan et al., 2020), domain-
specific terms for estate (Parlar et al., 2019), and
tourism (Arican et al., 2020). According to the
researchers behind the KeNet project, the lack of
necessary resources in the Turkish language and the
relatively well-structured KeNet program make it a vital
resource for further studies in this field [7].
Arabic WordNet (AWN): Limited Research and
Expanding Applications
Although Arabic is the official language for hundreds of
millions across 20 countries in the Middle East and
North Africa, it has seen minimal research in the fields
of computerization and lexical resources. Elkateb
(2005) emphasized that the richness of the Arabic
language necessitates the development of a lexical
resource such as an Arabic WordNet. Established in
2005, the Arabic WordNet (AWN) is based on the
widely recognized Princeton WordNet (PWN) in design
and content. AWN is directly mapped to PWN 2.0 and
EuroWordNet (EWN), enabling translation between
Arabic, English, and several other languages at the
lexical level.
Volume 05 Issue 12-2024
53
CURRENT RESEARCH JOURNAL OF PHILOLOGICAL SCIENCES
(ISSN
–
2767-3758)
VOLUME
05
ISSUE
12
Pages:
49-54
SJIF
I
MPACT
FACTOR
(2022:
6.
041
)
(2023:
7.
491
)
(2024:
8.235
)
OCLC
–
1242423883
Publisher:
Master Journals
Key criteria for selecting Sinsets for AWN include:
Connectivity: AWN must be densely interconnected
through chains like hyperonymy and hyponymy. Most
AWN Sinsets should correspond to English WN
analogs, and the general topology of the two networks
should be similar.
Relevance: Frequently encountered and salient
concepts are prioritized. Criteria include the frequency
of lexical elements in Arabic and English, as well as the
frequency of Arabic roots in relevant corpora.
Generality: Sinsets from the highest levels of the
WordNet hierarchy are given preference.
The construction of AWN occurs in two stages:
English to Arabic: For each Sinset in English, all
corresponding Arabic variants (if available) are
selected.
Arabic to English: For each given Arabic word, its
various senses are identified, and corresponding
English Sinsets are chosen.
AWN development also includes preparatory and
expansion tasks. AWN is not only aligned with PWN
(Fellbaum 1998) but also accommodates multi-
language support via interlingual indexes (ILI) or
ontologies, offering a multilingual user interface rather
than a merely bilingual one. Nevertheless, to
streamline the development and validation processes,
existing resources are leveraged as much as possible
[9].
UzWordNet: Expanding Uzbek into the Digital World
One of the projects aimed at integrating Uzbek into the
digital sphere and enhancing its IT applications is
UzWordNet. Created by Uzbek linguists and
developers, it serves as a corpus for the largest lexical
network available for the language. UzWordNet is the
initial WordNet-like resource for Uzbek, corresponding
to Princeton WordNet and encompassing 28,140
Sinsets, 64,389 senses, and 20,683 words [10]
The construction of UzWordNet is divided into three
stages
Selection and preprocessing of lexical resources.
Automated creation of a PWN-like structure for
Northern Uzbek (UzWordNet).
Human verification of the automated structure [10].
Challenges in Creating UzWordNet
Authors encountered multiple challenges during
UzWordNet’s development. Initially, the quality of
scanned dictionaries was poor, leading to errors when
converting electronic copies into text for analysis. This
issue was addressed by expanding each page visually
and processing it using free OCR (optical character
recognition) services. The dictionary was then
converted into a machine-readable format with a
tabular structure comprising three columns.
Like all word networks derived from PWN, UzWordNet
groups and classifies nouns, verbs, adjectives, and
adverbs into Sinsets based on lexical relationships. The
semantic tree structures of noun and verb Sinsets rely
on hyperonym-hyponym relations
Research by Alessandro Agostini, N. Abdurahmonova,
T. Usmanov, et al., titled "A Lexical-Semantic Database
for the Uzbek Language", presented initial results for
UzWordNet’s development
through expansion to
Northern Uzbek. The study’s evaluations achieved an
accuracy of 71.64% to 75.98% for over 280 sample
entries per part of speech (nouns, verbs, adjectives,
adverbs) [10].
CONCLUSION
The creation of thesaurus dictionaries laid the
groundwork for the development of WordNet
databases in global linguistics. WordNet opened a new
era in building extensive linguistic resources. Its
creation sparked interest in developing WordNets for
various national languages, contributing to significant
advancements in fields like machine translation,
computational lexicography, and corpus linguistics. It
also partly addressed translation challenges and
language diversity issues.
Volume 05 Issue 12-2024
54
CURRENT RESEARCH JOURNAL OF PHILOLOGICAL SCIENCES
(ISSN
–
2767-3758)
VOLUME
05
ISSUE
12
Pages:
49-54
SJIF
I
MPACT
FACTOR
(2022:
6.
041
)
(2023:
7.
491
)
(2024:
8.235
)
OCLC
–
1242423883
Publisher:
Master Journals
REFERENCES
1.
Matlatipov S., Abdurahmonova N. Modeling
Wordnet type thesaurus for uzbek language
semantik dictionary International Journal of
Systems
Engineering
2018;
2(1):
26-28)
http://www.sciencepublishinggroup.com/j/ijse
2.
Fellbaum Ch. A Semantic Network of English Verbs.
–
In: Fellbaum, C (ed) WordNet
–
An Electronic
Lexical Database.
–
The MIT Press. 1998.
–
pp. 69-
104.
3.
Abdurahmonova N. Muhammad Haydarov On the
tasks of creating a wordnet in uzbek language
“Ўзбекистонда
хорижий
тиллар”
илмий
-
методик электрон журнал journal.fledu.uz №
4/2019
4.
Robert M. Losee Decisions in thesaurus
construction and use Informating Processsing $
Management Volume 43, issue 4. July 2007. Pages
958-968
5.
Abjalova M. WordNet
–
lingvistik ontologiyalar
uchun tayanch baza. O‘zbekiston milliy universiteti
xabarlari. Toshkent
–
2023. №1.
6.
Bakay O.,
Ergelen O., Sarmıs E. Turkish WordNet
KeNet. January, 2021 Conference: Global Wordnet
Conference.
7.
Abjalova M. O‘zbek tili ontologiyasi: yaratish
texnologiyasi va konsepsiyasi. Monografiya.
Toshkent
–
2022.
–
B.54.
8.
Ehsani R. Kenet: a comprehensive turkish wordnet
and its applications in text clustering, isik university
–
2018.
9.
William B., Elkateb S. Introducing the Arabic
WordNet
Project
fellbaum-alkhalifa-206.pdf
(upc.edu)
10.
Alessandro A., Abdurahmonova N. va boshqalar “A
Lexical-Semantic Database for the Uzbek
Language”. 11th International Global Wordnet
Conference. January
–
2021.
