ENHANCING CROSS-LANGUAGE COMMUNICATION: A DEEP LEARNING APPROACH FOR INDIC LANGUAGE TRANSLITERATION

Abstract

This study aims to explore and understand the role of transliteration as a means of remote communication in India, with an emphasis on improving transliteration accuracy by employing an ensemble model of sequence-to-sequence with RNN layers as well as the attention mechanisms. The study suggests a new method for transliteration by integrating two models, each utilizing sequence-to-sequence with distinct structures: Long Short-Term Memory (LSTM) with attention and Gated Recurrent Unit (GRU) with attention. The models undergo training on a dataset of transliterated text and then put together through an ensemble technique to improve performance significantly. The efficiency of the ensemble model is determined using the Bilingual Evaluation Understudy (BLEU) score, a widely used metric for evaluating the quality of machine-translated text. The study employs three Indian languages, specifically Hindi, Marathi, and Bengali, to assess the effectiveness of the ensemble model in a Multi-Language Transliteration Ensemble task. The study offers a comprehensive analysis of the integration ensemble model, investigating its impact on transliteration accuracy, the individual contributions of each model to the ensemble, and the performance improvement resulting from the integration. The study's findings indicate that the ensemble model surpasses the individual models in terms of transliteration accuracy, emphasizing the potential of ensemble learning to enhance transliteration performance. This study makes a valuable contribution to the field of Natural Language Processing (NLP) by introducing a new ensemble model for transliteration that is capable of handling multiple Indian languages. Additionally, it provides a thorough analysis of the integration of the ensemble model. The results of this study have consequences for the creation of transliteration systems that can aid in long-distance communication and language acquisition in multilingual societies such as India.