Evolving questions in text planning

Abstract

This paper explores how the main question addressed in Text Planning has evolved over the last twenty years. Earlier approaches to text planning asked the question: How do we write a good text?, and whatever answers were found were programmed directly into code. With the introduction of search-based text planning in recent years, the focus shifted to the evaluation function, and thus the question became: How do we tell if a text is good? This paper will explore these evolving questions, and subsequent refinements of them as the field matures.