Final Report Summary - LOWLANDS (Parsing low-resource languages and domains)
Natural language processing (NLP) is an integral part of artificial intelligence, rapidly changing our daily lives. NLP, however, is not equally available across languages and domains, introducing what is commonly referred to as the digital language divide. Information and automated procedures are often not available to speakers of so-called low-resource languages. LOWLANDS (Parsing Low-resource Languages and Domains) aimed to push research horizons in cross language and domain adaptation of NLP models. We have developed a range of new learning algorithms and techniques for rapidly developing or transferring NLP models for low-resource languages, as well as increased interest in this research problem in the AI community.