Computational Plot Planning: A Temporal Social Network Approach
Computational plot planning is an important topic and challenge in the field of articial intelligence and computational creativity. Current methods dealing with the problem of story plot planning face difficulties in supporting long storylines and complex social structures. To solve this, we propose a new approach: using a temporal social network as a reference to plan future social interactions among characters in the developing plot. In this study, we analyze the development of social networks in stories through time and make predictions on future social network structures. Furthermore, we address three special properties of story social networks that each bring unique problems with them: newly introduced characters at each chapter, characters leaving the story, and special narrative focus. We used a LSTM recurrent neural network model to analyze four story data sets: the Odyssey, Iliad, Lord of the Rings and Game of Thrones. Based on this, our model attempts to predict a social network structure that fits the structure of the existing story. The model successfully predicted story social network structures that were similar to those in the original stories and shows promising potential for the plot planning field.
Haoran Ding, "Computational Plot Planning: A Temporal Social Network Approach", Master's Thesis for the Media Technology programme, Leiden University (The Netherlands), 2017