The Sense and Sensibility of Different Sliding Windows in Constructing Co-occurrence Networks from Literature

In this paper, we explore the design and effects of applying different sliding window methodologies to capture character co-occurrences within literature in order to build social networks. In particular, we focus our analysis on several works of 19th century fiction by Jane Austen and Charles Dickens. We define three different sliding window techniques that can be applied: collinear, coplanar, and combination. Through simple statistical analysis of each novel’s underlying textual properties we derive tailored window sizes for each case. We find that the selection of such parameters can significantly affect the underlying structure of the resulting networks, demonstrated through the application of different social network metrics on each of our novels. We also examine how the choice of window strategy can help address specific problems in current critical understanding of the novel.