Every page is one vector over the whole Marvel vocabulary: raw counts, counts without stopwords, or TF-IDF weights (stopwords kept, TF = count / page length, IDF = ln(303 / DF)); similarity is the cosine between two vectors. Words behind it are the terms that contribute most to that cosine. Linked means one page links to the other in the week 1 network. Click a name to jump to that character.
The most similar pairs of pages in the whole corpus that do not link to each other in either direction. Click a name to see its neighbors.