Data Efficient Machine Learning for Educational Content Creation

Machine Learning has revolutionized education by offering numerous practical applications. One such application is Neural Machine Translation (NMT) systems in the field of education, which hold immense social importance. These systems have the potential to make information accessible to diverse users in multilingual societies. By effectively translating audio, video, and textual content into vernacular languages, NMT systems greatly assist both students and teachers. However, when it comes to translating higher education or technical textbooks and courses, it becomes crucial for MT systems to adhere closely to the specific lexicon of the source and target domains. In this tutorial, we present our approach and framework to enable domain-aware translation without the need for parallel domain corpus. We will demonstrate several use-cases and applications that is widely used by hundreds of translators. We will also present our post-editing tool that assists translators in quickly correcting the machine translated text and reduce the cognitive load of users.