Title: 2' Title: To design and develop few modules to augment Machine Translation and CLIR systems from Eng
12. Title To design and develop few modules to
augment Machine Translation and CLIR systems from
English to Indian languages
Proposer Prof. Aditya Shastri Institution B
anasthali Vidyapith Rajasthan-304
022 Language/ English-Hindi Language pair
(and other North Indian Languages)
design
2Language/Language pair English-Hindi Name of
the component A Sense Disambiguator
for English-Hindi Translator (for the
source language) Technique(s) that will be
used (i) It will be based on the contextual
ranking of word Senses, and to have a better
Sense Disambiguation, it will also include
Synonyms of words that best suit the
context. (ii) Development of a phrasal term
English-Hindi dictionary, the use of which in
the disambiguator shall eliminate the
disambiguation phase for translation of phrases.
3Examples of disambiguation English If not
disambiguated 1. Status of well in
India - kegbÀDee/Deíe 2. Playing with the
ball - ieWo/vel³e 3. Rohan drank the
port - Menj/Heervee Phrasal terms dictionary
in MT 1. get rid of - ígìkeÀeje
Heevee 2. tit for tat - pewmes keÀes
lewmee 3. Raining cats and dogs - cetmeueeOeej
Jeltee
4Performance of these techniques in other
languages Phrasal term dictionaries are
being used on test basis in CLIR systems in
European languages where use of these have given
better quality in query translation as compared
to MT approaches. Domain for which the
performance will be optimized The
performance will be tested and optimised for
simple sentence translation rather than full text
translation The research will result
into - Component module for sense
disambiguator - A phrasal term dictionary as
lexical resource
5- Lexical Resource Phrasal term dictionary
- Language/Language It will be bilingual
(English- - pair Hindi) system.
- The work involves two phases
- development of a bidirectional phrasal term
dictionary (integration with usage of already
developed, if any). - - study of performance of both the MT and CLIR
systems with use of phrasal term dictionary.
6- Phrasal term dictionary
- In the oxford dictionary of about 2 lakh English
words - about 10 are phrases.
- The size of proposed dictionary - 20,000
phrases - Average size of French English - 75,000
- English-French - 40,000
- Technology to be used Inverted term list with
root word in English - Evaluation metric Improvement in performance
of the MT/CLIR system (as per defined
in part (i) of our work) with use of sense
disambiguator and phrasal term
dictionary
7Example phrases (Source Oxford advanced learner
dictionary) raining cats and dogs
(VP) cetmeueeOeej Jeltee pick a lock (V)
leeuee leesævee fall short (V) DeHe³eeHle
nesvee (root word underlined)
8PERT Chart for the entire project
II Design of disambiguator
_at_ 1000 phrase per month
Design of phrasal term dictionary inverted list
Addition of more data
Addition of more data
Integration testing
Testing with MT systems
Improvement Performance evaluation
Testing with CLIR system
Design of sense disambiguator
Development of sense disambiguator
6
12
18
24
30
36
I Evaluation metric
Design of test suit lexical
Design of test suit
Design of test suit
Design of test suit syntactic/ semantics
White box testing of MT systems
Activities
Design of CASE tool
Development of case tool
Black box testing Lexical/syntax
Black box testing semantic
18
30
36
24
6
12
Months