Capabilities and Past Work IIIT Hyderabad

1
Capabilities and Past WorkIIIT Hyderabad

2
Team

• Rajeev Sangal, PhD(CS), Prof.
• 24 years direct experience - Can devote 15 of
  time
• Dipti M Sarma, PhD(Ling.), Assoc. Prof
• 8 11 years - 50
• Soma Paul, PhD(Ling.), Research Asst. Prof.
• 2 years - 50
• Lakshmi Bai, PhD(Ling.), Adj. Prof.
• 4 30 years -- 30

3
Past Work Done

• Rajeev Sangal
• Indian language (IL) to IL Anusaarakas
• Morph analyzer (IL)
• Shallow parsing (IL)
• Parsing (IL)
• Paninian grammar framework (IL)
• Parsing (E)
• Word sense disambiguation (E)
• Parallel corpora (E-Hindi)

4
... Past Work Done

• Dipti M Sharma
• Bilingual dictionary (E-H)
• Linguistic tagging scheme (IL)
• Part-of-speech (POS) tagging scheme
• Chunking scheme
• Tree banking scheme
• Transfer grammar (E-IL)
• Paninian grammar framework (IL)

5
... Past Work Done

• Soma Paul
• Linguistic tagging and chunking
• POS tagging (Bengali)
• Chunking (Bengali)
• Phrasal verb lexicon (E)
• Phrasal verb recognizer
• HPSG framework (E)

6
... Past Work Done

• Lakshmi Bai
• POS tagging and chunking scheme (IL)
• Treebank scheme (IL)
• Sentence generation (IL)
• Syntax of IL (H, Telugu, Tamil)

7
Modules Developed

• Morph analyzer (H) gt95
• POS tagger (H) 93
• Chunker (H) 91
• Bilingual dictionaries
• E-H 28,000 words
• E-Marathi 20,000
• Marathi-H 20,000 (Can be used after
  refinement)
• Bengali-H 20,000 (Can be used after
  refinement)

8
... Modules Developed

• Sentence generator (H)
• Word generator (H)
• Transfer grammar (E-H) 150 rules

9
Tools Developed

• Morph. learning package
• Tagging interface To facilitate manual tagging
• POS tagging
• Chunking
• Tree bank development
• HMM training facilitator
• Blackboard architecture
• for system integration
• MT evaluation interface For human evaluation