Titlebook: Natural Language Generation; Ehud Reiter Textbook 2025 The Editor(s) (if applicable) and The Author(s), under exclusive license to Springe

檢查

敵手

Constrain

Rule-Based NLG,generated texts. Rule-based NLG has been overshadowed by neural NLG in recent years, but it is still the best way to build some NLG applications. Rule-based NLG also shows the types of decisions which need to be made in text generation. A good understanding of this helps anyone working in NLG, even if they use other approaches.

Mawkish

Requirements,looks at some of the different quality criteria that people may care about, workflows for using NLG (including human-in-the-loop), textual vs. graphical presentation of information, and methodologies for understanding (acquiring) requirements.

斥責(zé)

Evaluation,that are rigorous and replicable. This chapter discusses basic evaluation concepts and then describes techniques for human and automatic (metric) evaluation. It also looks at evaluating real-world impact of NLG systems and commercial evaluation.

最低點

Applications,cusses some fundamental issues (such as scalability) and then looks in more detail at four areas which NLG has been used commercially for many years: journalism, business intelligence, summarisation, and medicine. Lessons from these long-standing NLG use cases can be applied to newer applications of NLG.

arthroscopy

SEMI

口音在加重

Machine Learning and Neural NLG,This chapter gives an overview of machine learning and neural approaches to NLG, including language models. Because this area is changing very rapidly, the focus is on giving a high-level overview of basic concepts behind models and then discussing data and other issues which are important regardless of the model used.

CRACK

Textbook 2025pments have captivated both academic and commercial sectors, the focus has predominantly been on the latest innovations, often overlooking the rich history and foundational work in NLG. This book aims to provide a comprehensive overview of NLG, encompassing not only language models but also alternat