The very secret of the success of the Academy is its research since inception. It is conducting research in Sanskrit Speech Synthesis, Natural Language Processing, Machine Translation, Sanskrit teaching through computer media etc., for the last 8 years. This is one of the proofs of diversification from traditional research to scientific research through computers. It is also busy in surging into the field of Information Technology (IT). Besides, it is also collecting scientific information available in Sanskrit texts. It has also undertaken several science oriented projects sponsored by some well known organizations such as Indian Space Research Organization, HRD etc.

Blazing A Trail

Since Melkote being the place where Ramanujacharya resided for 12 years and propagated the Visistadvaita philosophy, naturally most of the research work in the Academy centers around Vedanta and its relevance to the present. The broad themes for research are:

Under each theme, projects are drawn up and a team of scholars work on the project. The projects, which have thus been undertaken by the Academy, are:

• VISISTADVAITAKOSA
  
  
• CRITICAL EDITION OF RAMANUJA GRANTHAMALA
  
  
• CRITICAL EDITION OF FOURTEEN MAJOR UPANISHADS
  
  
• BIBLIOGRAPHIES AND GLOSSARIES
  
  
• DRAVIDA - VEDANTA GRANTHAMALA - BHAGAVADVISAYAM

Research (Scientific)

KRIYA - Vibrantly Active

Since the late sixties, Artificial Intelligence (AI) researchers have been trying to build computer systems that can perform Natural Language Processing (NLP); but idioms, rhetoric, ambiguities and other idiosyncrasies that abound in natural languages have proved to be a formidable obstacle.

On the other hand, Sanskrit linguists, grammarians and philosophers of olden days have addressed many of these complex issues involved in the understanding of natural language. To examine the possibility of seeking solutions for these problems, a conference on Knowledge Representation in Sanskrit (KRIS) was held in Bangalore in 1986.

A paper presented by Rick Briggs, an AI researcher in US, observed "certain problems of concern in NLP have either been thought about or actually solved in the old tradition of Sanskrit grammar and philosophy..... If the grammatical view is not examined, current AI work will go on unnecessarily re - inventing a wheel millennia old".

This catalyzed the Academy to take up the Kriya Research Project which has the sanction of Department of Electronics, Government of India. It has been undertaken to sharpen our understanding of the essential features of the grammatical system employed in Sanskrit language under the guidance of senior Vyakarana vidvans. While working out procedures for the software, the systems analysts depend heavily on the shastras seeking the guiding principles for preparing the software. The II phase of this project called Sabdabodha was taken up in 1995.

NLP, whether the process is for Machine Translation or for other modes of handling language, has to begin with the parsing of sentences i.e. identifying, classifying and interconnecting the words in a sentence. Hence the Academy developed modules for identification, generation of different forms and analysis of verbs, participles and nouns in Sanskrit.

The following are the software that are being developed in this connection. These software's use the GIST technology developed by C- DAC for multi - Indian scripts.

1. Bodha

Developing this software is the crux of the Sabdabodha Project. The technique of disambiguating the meaning of a sentence according to the Navina Nyaya system is called the Sabdabodha.

At present the program is capable of parsing a valid sentence and identifying a factually incorrect sentence. Efforts are on to make this more comprehensive, at first by using a user - interface and getting rid of this later on.

2. Semusi

This module consists of Noun (Subanta) Generator and analyser. Having a lexicon consisting of words in Amarakosa, the generator is capable of generating forms in different cases and numbers (or any specific form) of the given word.

The Analyser identifies the ending (anta), the gender (linga) and base (pratipadika), Case (Vibhakti) and number (vachana) of the given noun. It also displays multiple identifications.

3. Prajna

This module consists of Verb (Tinanta) Generator and Analyser. With the generator, it is possible for the computer to generate all forms or any specified form of the given verb root. The Analyser identifies the Conjugation (Gana), Padi, Transitivity (Karma), It, Mode (Dhatuvidha), Voice (Prayoga), Tense (Lakara), Person(Purusa) and number (Vacana) of the given verb root. Multiple identifications are also displayed.

4. Cetana

This module consists of Participle (Krdanta) Generator and Analyser. The Generator is capable of generating 11 types of Krdanta forms, in 7+1 cases, three numbers and three genders. The Analyser identifies any form of the given Krd and gives the root, meaning (artha), Mode (Dhatuvidha), deciderative root (sanidhatu), Conjugation (Gana), Padi Transitivity (Karma), It, type of Krdanta, suffix (pratyaya), ending (anta), gender(linga), base(pratipadika), case (vibhakti) and number (Vacana).

5. Panini

This module combines two words as per Paninian rules handling exceptions carefully. It gives the effected sandhi type and also the Paninian rule governing it. It is a rule - based software.

The project has given raise to some spin - off software modules which are as follows:

a). Janani

It is a synonym retriever in Sanskrit which also provides etymology (vyutpatti), grammar (Nispatti) and usage (prayoga). It has a sufficiently large database based on the Amarakosa, Nirukta and Panini's Astadhyayi. Its merit lies in its power to compile and provide all information just at the press of a key.

b). Medha

It is a multilingual electronic dictionary versatile in several languages - Sanskrit, Hindi, Kannada, Telugu, Tamil, Malayalam and English. Any of these can be chosen as the source language and the other six automatically become the target languages. This is under preparation.

c). Bharati

This is a completely rule - based software. Based on Vrttaratnakara, it gives the metric analysis of any stanza, (at present only in samavrtta) with the number of laghus and gurus and their triplets (ganas), Chandas, Vrtta and the definition (laksana) for the particular metre.

d). Sasanam

It is a trilingual electronic law dictionary, the languages handled being English, Hindi and Kannada.

e). Nrpatunga

This is a contribution of the Academy to the Academy to the field of Kannada. Based on Paninian method, this module generates the different forms of verbs in Kannada. It handles two modes ordinary verbal form voices, two speeches, tenses and modes, assertive and interrogative, affirmative and negative, etc.

f). Pulavar

It is a synonym - retriever in Tamil based on Namadipanighantu.