Sinhala Ontology Generator for English to Sinhala Machine Translation

Abstract

Generating Ontology for a given word is a research challenging task that requires morphology, syntax, semantics and other linguistic knowledge. This paper presents a multi-agent system, named Sinhala Ontology Generator which is capable to generate ontology for a given Sinhala word. Ontology of the Sinhala word has been categorized by using 29 Sinhala part of speech tags including Noun, Verb, Adjective and Adverb. To generate Ontology for the given Sinhala word(s) Morphological, Grammatical, Semantic and Special attributes have been considered. The Ontology generator has been designed through the java based multi-agent system development framework named MaSMT. Sinhala Ontological generator communicates with Sinhala Morphological analyser, Sinhala syntax analyser, dictionary reader and word reader to collect information to generate the Ontology. The system has been tested with 400 sample Sinhala words. Experimental result indicates that morphologically and syntactically identified words give more than 70% of accuracy and unseen words give less than 50% of accuracy.