Computational Drafting of Plot Structures for Russian Folk Tales

Pablo Gervás

doi:10.1007/s12559-015-9338-8

Abstract

The plots of stories are known to follow general patterns in terms of their overall structure. This was the basic tenet of structuralist approaches to narratology. Vladimir Propp proposed a procedure for the generation of new tales based on his semi-formal description of the structure of Russian folk tales. This is one of the first existing instances of a creative process described procedurally. The present paper revisits Propp’s morphology to build a system that generates instances of Russian folk tales. Propp’s view of the folk tale as a rigid sequence of character functions is employed as a plot driver, and some issues that Propp declared relevant but did not explore in detail—such as long-range dependencies between functions or the importance of endings—are given computational shape in the context of a broader architecture that captures all the aspects discussed by Propp. A set of simple evaluation metrics for the resulting outputs is defined inspired on Propp’s formalism. The potential of the resulting system for providing a creative story generation system is discussed, and possible lines of future work are discussed.

Highlights

The concept of plot of a story is a useful abstraction
If the choice for a character function such as liquidation of misfortune or lack depends on which particular story action was chosen to instantiate the character function for lack, this procedure will both block non-appropriate instantiations for liquidation and will ensure the appropriate assignment of variable names to ensure coherence
The key measure to consider is, given a certain character function appearing in a candidate plot driver, how many of the functions preceding/following it in the plot driver are contained in the part of the reference sequence that goes before/after its appearances in the reference sequence

Summary

Introduction

The concept of plot of a story is a useful abstraction. It refers to the skeleton of the story, its elementary structure, and how the material in it all comes together into a single, coherentThe construction of stories is a cognitive task that is fundamental to the way humans understand the world and attempt to influence it. Pioneering work on automated storytelling dating back to the beginnings of artificial intelligence as a discipline had already considered dynamic modelling of the processes involved in creating stories [5, 15, 18, 19]. These efforts focused on making the most of available computational techniques—such as logic, planning, or case-based reasoning—to obtain story-like outputs. The idea to exploit structuralists account of narrative in computational systems

Methods

Results

Discussion

Conclusion