Interactive Natural Language Processing: Language Model as Agent - Wang, Zekun; Zhang, Ge; Lin, Chenghua

This book addresses the emergence of Interactive Natural Language Processing (iNLP) and its novel paradigm within the field of NLP and discusses the limitations in existing frameworks while aligning with the ultimate goals of artificial intelligence. This paradigm considers language models as agents capable of observing, acting, and receiving feedback iteratively from external entities.  The book provides a comprehensive survey of iNLP, starting by proposing a unified definition and framework of the concept. In addition, it provides a systematic classification of iNLP and dissects its various components, including interactive objects, interaction interfaces, and interaction methods.  The book then proceeds to delve into the evaluation methodologies used in the field and explores its diverse applications, scrutinizes its ethical and safety issues, and discusses prospective research directions. This work aims to be an entry point for researchers who are interested in this rapidly evolving area and offers a broad view of the current landscape and future trajectory of iNLP.