Individuação de autoria e identificação de estilo: análise de obras literárias com auxílio do R

Abstract

This paper adds to the works available on Natural Language Processing by providing a demonstration of how programming languages such as R (R CORE TEAM, 2020) can be useful in detecting authorship and identifying the style of the author in literary works. Two authors and two works each were selected, namely: The Adventures of Tom Sawyer (1876) and Adventures of Huckleberry Finn (1884) by author Mark Twain (1835-1910), and Typee: A Peep at Polynesian Life (1846) and Omoo: A Narrative of Adventures in the South Seas (1847) by author Herman Melville (1819-1891). Subsequently, the data were analyzed following the same methodology as Eder et al. (2016), in order to test the effectiveness of the stylo package and apply the Principal Component Analysis, Cluster Analysis and Consensus Tree methods. The results showed that each of the tested methods was able to distinguish the works of the authors, thus evidencing the effectiveness of the package used. In addition, a stylometric analysis is performed based on Craig's Zeta and Rolling Delta methods. For the latter, works by two German-speaking authors, Frank Kafka and Heinrich von Kleist, were used. The results pointed to a stylistic similarity of von Kleist, especially in Kafka’s first work. Additionally, Rolling Delta was used to explore an analysis carried by Juola (2013a, 2013b) regarding a work by J. K. Rowling written under the pseudonym of Robert Galbraith.