Resource Creation for SanskritHindi, Maithili Automatic Translation

1
Resource Creation for Sanskrit-Hindi, Maithili
Automatic Translation

• Project Proposal
• Submitted to
• The Department of Information Technology
• Ministry of Communication Information
  Technology,
• Govt. of India
• by
• Dr. Girish Nath Jha
• Special Center for Sanskrit Studies
• Jawaharlal Nehru University, New Delhi-110067.

2
Summary

• Title of Project
• Resource Creation for Sanskrit-Hindi,
  Maithili Automatic Translation
• Proposer
• Girish Nath Jha, Asstt. Professor,
• Institution
• Jawaharlal Nehru University, New Delhi 110
  067
• Language/Language pair
• Sanskrit-Hindi, Maithili
• Nature of Project
• Application oriented Research, Design and
  Development having production potential.

3
components to be implemented

• Sanskrit Karaka analyzer
• Sanskrit POS Tagger
• Annotated corpora for Sanskrit
• Multilingual online Amarakosha
• Lexical resources for Sanskrit-Hindi
• Lexical resources for Sanskrit-Maithili
• Text generator for Hindi
• Text generator for Maithili
• Paradigm sets for Maithili
• Paradigm sets for Hindi

4
techniques to be used

• Paninian Karaka techniques for analysis
• Sanskrit lexicography and morphological
  approaches for POS Tagging
• Ontological lexicography (Amarakosha) and word
  net approaches
• Lexical resources creation techniques for
  Sanskrit, Hindi and Maithili
• Paradigm Text generation techniques for Hindi and
  Maithili

5
Previous Project experience

• Online Multilingual Amarakosa (2005-) As
  Director funded by JNU under UPOE program of
  UGC
• Database system for dialects of Hindi
  (2002-2003) as co-director
• ALED (Active Living Everyday) 2001-2002 at Human
  Kinetics, IL, USA
• BISS Databse develoment project at the University
  of Illinoiss Beckman Institute for Advanced
  Science and Technology 1998-1999
• Analyzing schizophrenic speech at Dept. of
  Psychology, University Of Illinois,
  Urbana-Champaign, 1997
• Developing technical terminology for Hindi
  (participated in a series of workshops organized
  by Scientific and Technical Terminology
  Commission, Min of HRD, 1994-1996) and
  contributed in developing technical terminology
  for Hindi
• Computer Assisted Sanskrit Teaching Learning
  Environment (DoE sponsored project at JNU
  1991-1994) research investigator

6
workforce

• JNU is a leading university with strong
  multidisciplinary approach
• Large number of students from the linguistic and
  Sanskrit centers are taking courses in
  computational linguistics and are available for
  project work
• 4 Ph.D students
• 4 M.Phil students
• 30 MA students

7
performance of these techniques in other languages

• The Paninian Karaka approach has been used in
  Anusaaraka (Sanskrit-English)
• Morphological analysis approaches are generally
  used in POS tagging for Indian languages
• Ontological principles are embedded in the word
  net approaches (e.g. for Hindi by IIT Mumbai)
• Lexical and data oriented approaches are used in
  most statistical or hybrid systems (e.g. Shakti
  system of IIIT)
• Paradigm and text generation techniques have been
  used in AnglaBharati

8
work done so far

• The English-Hindi lexical semantic component
• Sanskrit Subanta generator
• English SQL NLI
• Online Multilingual Amarakosha
• Sanskrit Analysis System
• Sandhi analyzer
• Subanta analyzer
• Dhatu database
• POS tagging
• Tagged MW Dictionary
• Karaka analyzer
• E-corpora (sandhi free, with sandhi)
• Sample Online lexicon based tagger
• Sample online lexicon based translator