Intertextuality detection in literary texts using Word2Vec models

This paper presents a research on using Google's Word2Vec model for intertextuality detection in literary texts. Word2Vec is considered a new and powerful Natural Language Processing semantic tool, used for computing text cohesion and similarity between documents. This article analyzes intertextuality in terms of semantic relations between words in novels. A similarity is calculated as a score: the higher it is, the stronger the relatedness between those two parts of text is. For experiments, different windows of text are used considering the dimensions of the two texts analyzed. For the highest scores matchings, a graphical representation is made to illustrate the parts of text which are linked.