networktext

a tool for creating character co-occurrence networks from a text (book, etc).

what it does now:

• reads text and epub files
• recognizes entities
  • things are cached so they are quick (ish) and only reload when the entities change
• allows for annotations of:
  • entity disambiguation (pseudonyms/aliases)
  • file restriction (i.e., don't include this section of the epub)
  • file ordering (read the network in this way)

this isn't well documented or ordered, nor is it tested

what it should do in the future:

• support a commandline interface
• allow for scoping of entities
• give you the sentences that an entity occurred in
• be smart about listing entities
  • group them by shared substrings
  • show you the longest first (eg, if there's one character named Adam Smith who appears as Adam and Adam Smith, show Adam Smith first)
  • case insensitive?
• auto create the key?