Translation Based

Translation-based research focuses on improving the accuracy and fluency of machine translation (MT) systems, particularly for low-resource languages and challenging linguistic phenomena like idioms and proper names. Current research employs transformer-based neural networks and explores techniques like matrix completion for lexicon induction, data augmentation using storyboards, and parameter-efficient fine-tuning to address data scarcity. These advancements aim to enhance cross-lingual understanding, improve the accessibility of information across languages, and enable more effective applications in diverse fields such as healthcare (e.g., simplifying prescription instructions) and cross-lingual information retrieval.