Character-based kernels for novelistic plot structure

Proceedings of the 13th Conference of the European Chapter of the Association for Computational Linguistics, pages 634–644, Avignon, France, April 23 - 27 2012. c⃝2012 Association for Computational Linguistics Character-based Kernels for Novelistic Plot Structure Micha Elsner Institute for Language, Cognition and Computation (ILCC) School of Informatics University of Edinburgh melsner0@gmail.com Abstract Better representations of plot structure could greatly improve computational meth- ods for summarizing and generating sto- ries. Current representations lack abstrac- tion, focusing too closely on events. We present a kernel for comparing novelistic plots at a higher level, in terms of the cast of characters they depict and the so- cial relationships between them. Our kernel compares the characters of different nov- els to one another by measuring their fre- quency of occurrence over time and the descriptive and emotional language associ- ated with them. Given a corpus of 19th- century novels as training data, our method can accurately distinguish held-out novels in their original form from artificially dis- ordered or reversed surrogates, demonstrat- ing its ability to robustly represent impor- tant aspects of plot structure. 1 Introduction Every culture has stories, and storytelling is one of the key functions of human language. Yet while we have robust, flexible models for the structure of informative documents (for instance (Chen et al., 2009; Abu Jbara and Radev, 2011)), current approaches have difficulty representing the nar- rative structure of fictional stories. This causes problems for any task requiring us to model fiction, including summarization and generation of stories; Kazantseva and Szpakowicz (2010) show that state-of-the-art summarizers perform extremely poorly on short fictional texts1. A ma- jor problem with applying models for informative 1Apart from Kazantseva, we know of one other at- tempt to apply a modern summarizer to fiction, by the artist Jason Huff, using Microsoft Word 2008’s extrac- tive summary feature: http://jason-huff.com/ text to fiction is that the most important struc- ture underlying the narrative—its plot—occurs at a high level of abstraction, while the actual narra- tion is of a series of lower-level events. A short synopsis of Jane Austen’s novel Pride and Prejudice, for example, is that Elizabeth Ben- net first thinks Mr. Darcy is arrogant, but later grows to love him. But this is not stated straight- forwardly in the text; the reader must infer it from the behavior of the characters as they participate in various everyday scenes. In this paper, we present the plot kernel, a coarse-grained, but robust representation of nov- elistic plot structure. The kernel evaluates the similarity between two novels in terms of the characters and their relationships, constructing functional analogies between them. These are in- tended to correspond to the labelings produced by human literary critics when they write, for exam- ple, that Elizabeth Bennet and Emma Woodhouse are protagonists of their respective novels. By fo- cusing on which characters and relationships are important, rather than specifically how they inter- act, our system can abstract away from events and focus on more easily-captured notions of what makes a good story. …

Content truncated to 500 words