Sequence to sequence networks for Roman-Urdu to Urdu transliteration

Abstract

Neural Machine Translation models have replaced the conventional phrase based statistical translation methods since the former takes a generic, scalable, data-driven approach rather than relying on manual, hand-crafted features. The neural machine translation system is based on one neural network that is composed of two parts, one that is responsible for input language sentence and other part that handles the desired output language sentence. This model based on encoder-decoder architecture also takes as input the distributed representations of the source language which enriches the learnt dependencies and gives a warm start to the network. In this work, we transform Roman-Urdu to Urdu transliteration into sequence to sequence learning problem. To this end, we make the following contributions. We create the first ever parallel corpora of Roman-Urdu to Urdu, create the first ever distributed representation of Roman-Urdu and present the first neural machine translation model that transliterates text from Roman-Urdu to Urdu language. Our model has achieved the state-of-the-art results using BLEU as the evaluation metric. Precisely, our model is able to correctly predict sentences up to length 10 while achieving BLEU score of 48.6 on the test set. We are hopeful that our model and our results shall serve as the baseline for further work in the domain of neural machine translation for Roman-Urdu to Urdu using distributed representation.

RUTUT: Roman Urdu to Urdu Translator Based on Character Substitution Rules and Unicode Mapping

“…1736 distinct words are translated into English successfully out of 2000 which shows the accuracy is 86% [49]. Neural machine translation model is based on encoder-decoder architecture [32] using sequence to sequence learning methods. This model consists of two parts one takes the input sentence and the second is responsible for the output.…”

“…The neural network based on the sequence to sequence network model [32], [36], [51] has become a very successful and popular technique to predict the identical sequence for mapping purposes. The kind of problems like handwriting generation, conversational modeling, the secondary structure of protein prediction, question answering, text to speech, music, [52] modeling of polyphonic music, speech recognition, machine translation and modeling of the speech signals are solved by applying these neural network-based models.…”

“…Produce incorrect sentences linguistically. [32] Neural Network based Roman Urdu to Urdu Translator proposed with decoding and encoding techniques.…”

A Study of Neural Machine Translation from Chinese to Urdu

“…Urdu transliteration into sequence to sequence learning difficulty. The Urdu corpus was created and pass it to neural machine translation that guess sentences up to length 10 while achieving good BLEU score [23] . Neelam…”

Natural language understanding of map navigation queries in Roman Urdu by joint entity and intent determination

“…Roman Urdu has been popularized in the last 2 decades due to increased use of Urdu writing on the internet and mobile phones using the standard English keyboard. Though there are examples of deep learning applied to Roman Urdu text, those are in NLP domains other than natural language understanding, like sentiment analysis (Ghulam et al, 2018;Shakeel & Karim, 2020) and Roman Urdu to Urdu transliteration (Alam & ul Hussain, 2017). In this research, we are going to work on natural language understanding of a Roman Urdu navigational dialogue dataset.…”