Automatic construction of dictionary on English-Nepali parallel corpus

Abstract

This dissertation describes an approach based on word alignment on parallel corpora, which aims at facilitating the lexicographic work of dictionary building. The proposed model of dictionary construction first perform the tokenization of input text and then TnT tagger is used for tagging the tokenized text, after this the word alignment is done to find out the word pair form source (English)language and target (Nepali) language. Finally our dictionary generation algorithm generate the sample dictionary formed from the word that made the given input text. Our model does rely on the information from tagging as well. Hence the model accuracy not only depends on the alignment algorithms and the training corpus but also depends on the accuracy of tagger. This corpus-driven technique, in particular the exploitation of parallel corpora, proved to be helpful in the creation of bilingual dictionaries for several reasons. Most importantly, a parallel corpus of appropriate size guarantees that the most relevant translations are included in the dictionary.