Automatic Analysis of Plot for Story Rewriting

Automatic Analysis of Plot for Story Rewriting Harry Halpin School of Informatics University of Edinburgh 2 Buccleuch Place Edinburgh, EH8 9LW Scotland, UK H.Halpin@ed.ac.uk Johanna D. Moore School of Informatics University of Edinburgh 2 Buccleuch Place Edinburgh, EH8 9LW Scotland, UK J.Moore@ed.ac.uk Judy Robertson School of Informatics University of Edinburgh 2 Buccleuch Place Edinburgh EH8 9LW Scotland, UK judyr@inf.ed.ac.uk Abstract A method for automatic plot analysis of narrative texts that uses components of both traditional sym- bolic analysis of natural language and statistical machine-learning is presented for the story rewrit- ing task. In the story rewriting task, an exemplar story is read to the pupils and the pupils rewrite the story in their own words. This allows them to prac- tice language skills such as spelling, diction, and grammar without being stymied by content creation. Often the pupil improperly recalls the story. Our method of automatic plot analysis enables the tu- toring system to automatically analyze the student’s story for both general coherence and specific miss- ing events. 1 Introduction StoryStation is an intelligent tutoring system cre- ated to provide personalized attention and detailed feedback to children ages 10-12 on their writing (Roberston and Wiemar-Hastings, 2002). Writing is viewed as a skill-based task, with skills being elements of writing such as spelling, diction, and plot development. Each writing skill is associated with an animated agent that provides online help. Evaluations of StoryStation show that children en- joy the personalized encouragement and construc- tive comments that StoryStation provides (Robert- son and Cross, 2003). StoryStation was designed by researchers in conjunction with two teachers and a group of students. However, both students and teachers indicated StoryStation would be signifi- cantly improved if it were enhanced with an agent that could give feedback about the plot of a story. Here we describe how techniques from symbolic natural language processing and statistical machine- learning were used to tackle the problem of auto- mated plot analysis for StoryStation. 2 The Story Rewriting Task In the story rewriting task, pupils rewrite a story in their own words, allowing them to focus on their writing ability instead of plot formulation. This task is currently used in Scottish schools and thus it was chosen to be the first feature of the plot analysis agent. We collected a corpus of 103 stories rewritten by children from classes at primary schools in Scot- land. Pupils were told a story, an exemplar story, by a storyteller and were asked to rewrite the story in their own words.1 The automated plot analysis program must be able to give a general rating of the quality of the rewritten story’s plot and be able to determine missing or incorrect events. The general rating can be used by the teacher to find out which pupils are in need of attention, while the more spe- cific details can be used by an animated agent in StoryStation to remind the student of specific events and characters they have forgotten or misused. 3 Plot Ratings The stories were …

Content truncated to 500 words