Knowledge-augmented Methods for Natural Language Understanding

This chapter delves into the emerging domain of knowledge-augmented natural language understanding (NLU), an essential aspect of natural language processing. The integration of external knowledge sources with pretrained language models is key to tackling a wide range of NLU tasks, including question answering, fact verification, and knowledge graph-based tasks. This chapter systematically explores the three steps of knowledge integration—representation, grounding, and integration, with analysis of both structured and unstructured knowledge sources. Through an examination of state-of-the-art models and techniques, the chapter offers a close look at recent advances in knowledge-augmented NLU domain. This comprehensive overview aims to equip readers with a deeper understanding of the current landscape and future potential of knowledge-augmented NLU.