Application
摘要
This chapter presents a comprehensive exploration of the various applications of interactive natural language processing (iNLP). Section 11.1 delves into Controllable Text Generation (CTG), discussing how interaction with humans, knowledge bases, models, tools, and environments can enhance system customization to meet specific user constraints and preferences. Section 11.2 examines the role of iNLP in writing assistance, categorizing functionalities from basic content support to sophisticated content co-creation, and emphasizing the iterative nature of human-model text refinement. In Sect. 11.3, we investigate Embodied AI, where language models observe, manipulate, navigate, explore, and engage in multi-agent tasks within simulated or real-world environments. Section 11.4 focuses on Text Games, highlighting the significant potential of iNLP, in both playing text-only games and augmenting text-aided gaming experiences. Lastly, Sect. 11.5 discusses the broader applications of iNLP, including specialization and personalization of language models, as well as model-based evaluation techniques that align more closely with human evaluation. This chapter underscores the versatility of iNLP in enhancing the interactivity and utility of language models across diverse domains and applications.