What is the expected category of contribution: CLIR

1
Title Cross Lingual Information Retrieval for
Indian Languages
Proposer CDAC Noida
Name of the company CDAC Noida
Language/Language pair English, Hindi, Punjabi
and Marathi
What is the expected category of contribution
CLIR
2

• Strength of CDAC Noida
• Technical Capabilities NLP Lab equipped with
  necessary software
• and 50
  trained software professional
• Manpower Involved
• Technical
• K K Arora
• Vijay Kumar
• Manish Kumar
• Pragdeeshvaran
• Linguistic Support
• Prof K K Goswami
• Prof Thakur Das
• Mrs Ragini
• Prof Jagannathan, Mr Chandra Mohan and others

3
List previous collaboration with universities/RD
institutions

• IIT Kanpur CHD, Delhi CSIO, Chandigarh
• ISI Kolkata Kendriya Hindi Sansthan
• DRDO, Delhi Delhi Press Prakashan Pustak
  Mahal
• IISc Bangalore CSTT, New Delhi COCOSDA, Japan
• Sahitya Akademi IIT Roorkee ELDA, France
• MGAHV, Wardha Abbyy, Russia W3C
• Jamia-Milia GKV, Hardwar BITS Pilani
• GBPUAT, PantNagar Kumaon Univ, Nainital
  Banasthali Vidyapeeth

4

• Previous work done in this or similar areas
• Machine Translation
• Parallel Corpus for Indian languages
• Dictionaries / Terminologies like Shabdika,
  Lexicon for
• MAT, IT Terminology
• Prototype development for CLIR

5
Need of the CLIR Indian Context

• Availability of Content on Internet in multiple
  Indian languages
• People in India generally are familiar with more
  than one language
• To retrieve the related information that may be
  available in any of the known set of languages by
  querying in one of these languages.

6
Block diagram for CLIR
7
Sample entry in Dictionary
8
Language Resources needed CLIR

• Dictionary of words in English, Hindi, Punjabi
  and Marathi
• Root words dictionaries
• Database of phrases and collocations
• A stop word lists in Hindi, Marathi and Punjabi
• List of Hyphenated words
• Proper Names database

9
Language Tools needed CLIR

• Inflator/Stemmer routines for English, Hindi,
  Punjabi and Marathi
• Transliteration routine for language pairs of
  targeted languages. This will be based on hybrid
  approach of Rule based and dictionary based

• Inflator/Stemmer routines for English, Hindi,
  Punjabi and Marathi
• Transliteration routine for language pairs of
  targeted languages. This will be based on hybrid
  approach of Rule based and dictionary based
• N-gram Analyzer tool
• Spell variant generator
• Proper name extractor from Parallel corpus

10

• Thank You