La comparaison de la lecture humaine avec la lecture machine dans le Système de la Nature de d’Holbach

In this empirical study, I explored thematic patterns in d’Holbach’s Système de la Nature. The overall purpose of my thesis was to compare the thematic structure of the corpus, in terms of co-occurrence and collocations of the 8 main themes, in human-reading and machine-reading. To achieve this, I used two methods: First, I used a qualitative data analysis software (NVivo 10) to do the comparative analysis of the main themes in consideration of examining the density of the appearance of each theme together. Next, I used RStudio Mallet software package for topic modeling to compare the co-occurrence of these main themes and to discover topics from the corpus. My results revealed that human-reading and machine-reading share only 12.5% of similarity in terms of semantically correlated themes. Additionally, topic models identified some marginalized themes associating with the philosophy of materialism, such as Cerveau, Nation and Vertu.