Evaluating the Effects of Natural Language Generation Techniques on Reader Satisfaction.pdfVIP

Evaluating the Effects of Natural Language Generation Techniques on Reader Satisfaction Charles B. Callaway (cbcallaw@eos.ncsu.edu) James C. Lester (lester@csc.ncsu.edu) The IntelliMedia Initiative Department of Computer Science North Carolina State University Raleigh, NC 27695 USA Abstract Zwaan 1997; Hoover 1997) have focused on mechanical aspects such as reading rate, and did not have access to We are witnessing the emergence of a new technology computational mechanisms for producing the texts they for dynamically creating stories tailored to the interests of particular readers. Narrative prose generators offer studied. much promise for literacy education, but designing them To study the changes in perceived text quality stem- for maximal effectiveness requires us to understand their ming from alterations to the underlying text generation effect on readers. This article describes the evaluation architecture, we conducted a formal study gauging the of STORYBOOK, an implemented narrative prose genera- tion system that produces original fairy tales in the Little satisfaction of subjects reading narratives. The study in- Red Riding Hood domain. STORYBOOK creates two to volved the following: three pages of text consistently represented at the deep linguistic structure level. Because of this, we can for- A consist