Machine Translation using Dictionary based Techniques

Abstract

Translating words and sentences is a major challenge, let alone a whole book. Tradition-ally, publishers used to manually translate a whole book. As computers and Internet came into being, it was used for translating words and sentences from one language to other using various methods. Dictionary based techniques are one of the oldest machine translation techniques. Mod-ern day apps use statistical machine translation techniques along with neural networks and/or natural language processing methods, where there is a high chance of error as they focus more on the language diversity. The major mo-tivation of this research work is to improve the English learning skills of ru-ral South Indian school students. As per a survey conducted by the authors, the rural students lack in command over English, due to the difficulty in translation services. The authors created a dataset consisting of English words from Tamil Nadu State textbooks of classes up to grade 10, to execute the translation. In this paper, we have used dictionary mapping, such as lin-ear, binary and Trie search methods, for performing word translation from English to Tamil, and compared them in different conditions to identify the best one. It is observed that the Binary search method performs better in all cases and hence it was selected to be implemented in the translation app. It is then clubbed with natural lan-guage processing (NLP) techniques and Rule Based Machine Translation (RBMT) techniques to carry out the trans-lation of a whole sentence. This whole technique is integrat-ed into an app, which is intended for students who are not fluent in English and requires an assistance.