antonymy tool ppt - cfilt

Download Report

Transcript antonymy tool ppt - cfilt

Antonym Creation Tool
Presented By
Thapar University WordNet Development Team
Introduction to WordNet

A WordNet is a large lexical database of a language.

Nouns, verbs, adjectives and adverbs are grouped into sets of
cognitive synonyms (synsets), each expressing a distinct
concept.

Each synset in WordNet is related to other synsets by well
known two types of relations.
-Semantic Relation
-Lexical Relation
Semantic Relation

Semantic relations are the relations between two whole synsets.

Same for all the languages.

For example Hpernymy/hyponymy relation , Consider two
synsets given in (1) and (2).

{ਪੌਦਾ paudā ’plant’, ਬੂਟਾ būṭā ’plant’}
… (1)

{ਚਾਹ cāh ‘tea’}
… (2)

Here {ਪੌਦਾ paudā ’plant’, ਬੂਟਾ būṭā ’plant’ } is hypernym of
{ਚਾਹ cāh ‘tea’} and {ਚਾਹ cāh ‘tea’} is hyponym of {ਪੌਦਾ
paudā ’plant’, ਬੂਟਾ būṭā ’plant’ }
Lexical Relation
•
Lexical relations are language dependent.
•
Lexical relations are the relations between members of two different
synsets.
•
For example consider two synsets given in (1) and (2)
{ਮੋਟਾ mōṭā ‘fat’, ਭਾਰੀ bhārī ‘fat’, ਸਥੂਲ sathūl ’fat’, ਥੂਲ thūl ‘fat’, ਨ ਿੱਗਰ
niggar ’fat’, ਵਜ ੀ vajnī ‘fat’}
…(1)
{ਪਤਲਾ patlā ’thin’, ਦੁਬਲਾ dublā thin,ਕਮਜ਼ੋਰ kamzōr ’thin’, ਮਾੜਾ māṛā
’thin’}
•
...(2)
Here, synsets (1) and (2) are opposites but they does not share antonym
relation. Antonym relation exist between two words not between two
synsets. Here the words ਮੋਟਾ mōṭā ’fat’ and ਪਤਲਾ
patlā ‘thin’ are in
antonym relation but not the words ਮੋਟਾ mōṭā ’fat’ and ਕਮਜ਼ੋਰ kamzōr ’thin’
Antonym Creation Tool

Helps to create the antonyms for any language.

Eliminates the manual typing effort.

Tool provides a user friendly interface for creation of antonyms.
Idea Behind Development of Antonym
Creation Tool

Antonyms for Hindi Language has been already created by IIT,
Bombay.

All Indian Language WordNets are being created using
expansion approach from Hindi WordNet.

So, antonyms created in Hindi Language will be covered in all
other languages.
Hindi WordNet Antonyms

Hindi WordNet antonyms are classified into four major categories
- Noun
- Verb
-Adjective
- Adverb

These are further classified into 12 subtypes
- Action
- Personality
- Amount
- Place
- Color
- Quality
- Direction
- Size
- Gender
- State
- Manner
- Time
Earlier Database Design for Storage of Antonyms
Noun
Verb
Adjective
Adverb
Contd..
tbl_anto_noun_action
Contd..
tbl_anto_verb_action
Contd..
tbl_anto_adjective_action
Contd..
tbl_anto_adverb_action
Challenges for Creation of Antonym Creation Tool

Since the WordNet database has been modified by Goa
University, So it is important to design a tool which can port
the data into new database from old database.
wn_rel_antonymy
Development Approach

To cover all the antonyms created in Hindi WordNet.
Description of Type
Description of Subtype
Description of Hindi Language Records
Contd..
tbl_anto_noun_action
Populate Records of Target Language Based on
Hindi Language Records
Contd..
wn_synset_word
wn_word
How to Save
Storage of Records in New Database

The Records are successfully saved in the database as shown
wn_rel_antonymy
wn_property_antonymy_ gradation
Contd..
wn_rel_antonymy
wn_property_antonymy_ gradation
Contd..
wn_rel_antonymy
wn_property_antonymy_ gradation
Installation of Antonym Creation Tool

The prerequisites for installation are following
-Java version 1.6
-MySql (Preferably Install Xampp Tool freely available on internet)
-Window 7
How to Install

Install JDK and MySql (preferably Xampp tool with integrated
apache support)

Install Sakal Bharti Font

After successful installation of
Xammp you will find Xampp
folder in c drive.

Open the data Folder using the following path

C:\xampp\mysql\data
Contd..

Copy iwn_web_unicode, wordnet_master and wordnet_target
databases in the above folder

Once you are ready with the databases you can open anto_tool.jar
with java and start working provided anto_tool.jar and lib folder
should be placed in same location.