Analyses of character networks in dramatic works by using graphs

Official URL: https://dx.doi.org/10.1109/BESC51023.2020.9348328

Artificial Literature (ALit) starts seem possible with upcoming generative models. ALit consists of writing machines that generates literary works. Although there are random machines that imitates the language models, texts by the writing machine should be far beyond, they need to have the structural similarity with the reference texts. In the framework for ALit, our first task is to find structure of tragedies which are very well stated beginning with Aristotle. In this piece of work, the character networks are analyzed with graph theory in order to extract structural properties of Shakespearean texts. The character network is generated and represented as undirected weighted graphs. The weighted and betweenness centrality graphs are interpreted with and without protagonists/antagonists following the 'Network Theory, Plot Analysis' by Franco Moretti. As a conclusion, we investigated symmetries or antagonism clusters. There is an antagonism behind the protago-nist/antagonists. This investigation is important to extract knowledge about the class or political struggle.