Synthetic Data Generation and Joint Learning for Robust Code-Mixed Translation

TL;DR

This paper introduces a synthetic Hinglish-English dataset and a joint-training model, RCMT, to improve code-mixed translation robustness, demonstrating superior performance and adaptability to low-resource and noisy scenarios.

Contribution

The paper presents HINMIX, a large synthetic Hinglish-English corpus, and RCMT, a novel robust joint-training model for code-mixed translation handling noise and low-resource challenges.

Findings

RCMT outperforms state-of-the-art methods in code-mixed translation.

HINMIX provides a large synthetic parallel corpus for Hinglish-English.

RCMT demonstrates zero-shot adaptability to Bengalish-English translation.

Abstract

The widespread online communication in a modern multilingual world has provided opportunities to blend more than one language (aka code-mixed language) in a single utterance. This has resulted a formidable challenge for the computational models due to the scarcity of annotated data and presence of noise. A potential solution to mitigate the data scarcity problem in low-resource setup is to leverage existing data in resource-rich language through translation. In this paper, we tackle the problem of code-mixed (Hinglish and Bengalish) to English machine translation. First, we synthetically develop HINMIX, a parallel corpus of Hinglish to English, with ~4.2M sentence pairs. Subsequently, we propose RCMT, a robust perturbation based joint-training model that learns to handle noise in the real-world code-mixed text by parameter sharing across clean and noisy words. Further, we show the adaptability of RCMT in a zero-shot setup for Bengalish to English translation. Our evaluation and comprehensive analyses qualitatively and quantitatively demonstrate the superiority of RCMT over state-of-the-art code-mixed and robust translation methods.