SENTENCE ANALYSIS USING PANINIAN APPROACH TO GRAMMAR

Abstract

This thesis is concerned with the Implementation of a natural language understanding package, using Sanskrit grammar (Karak charts) and conceptual graphs. The package converts imperative and exclamatory sentences in English to an intermediate representation, using Sanskrit. An active lexicon has been implemented, which stores hundreds of words together with their lexical attributes. A verb-centered parser uses this information to semantically analyse the input sentence. It generates output in the, form of a conceptual graph with Karak and non-karak relations between the words of a sentence. In earlier works (12,15] declarati,Yve and interrogative sentences in English were considered. It is also shown in the present work how these two packages can be combined with the present one to give a comprehensive package for natural language understanding. This package can also be used as the basis of a question - answering system in natural languge. The implementation of the system indicates that the Sanskrit based approach may be extended to other languages too. The implementation of the package is done in Turbo C on a PC-AT compatible.