Quantitative thematic diversification and evolution of classical science fiction in the public domain based on complex network analysis and natural language processing

Science Fiction (SF) is a young but thriving modern literary genre characterized by vivid portrayals of alternate worlds featuring advanced science and technology, often distant in space and time from ours. Establishing captivating, fantastical, and futuristic environments for imaginative storytelling constantly required profound imagination and innovative thinking of the genre’s creators, propelling the young literary genre to evolve and grow in a relatively short time into one that now exerts a strong influence even outside its original realm of literature, such as cinema and television. As creative works in the written form, the texts of SF novels are the primary source for understanding their nature and characteristics, including the developmental history. In this paper we study in detail how SF has evolved since its beginning via two quantitative techniques, computational linguistics (natural language processing) and network science, jointly applied to a comprehensive data of classical SF in the public domain. We find that the network constructed between the texts based on linguistic similarity enables us to detect the emergence and evolution of distinct themes across different generations, and that the most important events in the history of SF strongly correlate with moments of rapid growth in the genre’s thematic diversity. This shows the necessity of a continuous infusion of new ideas and the resulting elevation in diversity for the success and growth of a creative genre. Also, in this age of a strengthening interest in the scientific understanding of human creativity and machine intelligence, this work represents a contribution to one of the most promising yet underrepresented topics in human-centered data science.