Application of Scattered Context Grammar for Resolving Dependency in Hindi Language

Abstract

Parsing Indian languages has always been a challenging task. In recent years there have been various approaches explored for improving parsing accuracy for Hindi and other Indian languages. In this work, we present our experiments to improve dependency parsing for Hindi language. The Hindi language is considered amongst the richest morphological language because of the various forms generated from a single word form. In this paper, dependencies among various words in Hindi sentences are entertained. Networkx is used over here to generate the dependencies between the words. The dependency graph generated from the networkx clearly defines the dependencies between the words. These dependencies are converted to their plural forms using the rules to convert various singular word forms to their plurals. So this work also includes the rules which are to be considered while converting nouns, adjectives and verbs from their singular form to plural form using Transducers. The second part of the thesis includes the use of scattered context grammar which is used in representing the dependencies among the words. These dependencies are generated when nouns, verbs and adjectives are converted from singular to their plural forms. Scattered context grammar is used to represent these dependencies using its generalised production rules. This study also presents the production rules in which sentences are transformed on the basis of neither-nor clauses, Interrogative clauses, existential clauses, sentences having question tag and generation of grammatical sentences. So this paper also covers a step higher of scattered context grammar which is transformational scattered context grammar in this work.