Charnetto

This module is designed to create an automated character network based on a book or a movie script.

Getting started

Charnetto is implemented both with spaCy and Flair for the named entity recognition step. Please install the desired library using

pip install spacy

or

pip install flair

For more information, see the online documentation.

How to use charnetto

You can use the Jupyter Notebook charnetto_example.ipynb to see a full example of how to generate a character network based on a book (with Flair as a NER tool or with manual annotations) or on a movie script.

Supported data

• The books must be in .txt, ideally with one paragraph per line. For english books, replacing ,' by ', at the end of dialogues tends to give better results with the NER part.

• The movie scripts need to resemble those available on IMSDB: the regex for character names looks for capital letters preceded by at least two line breaks.

• If you want to annotate some books manually, you can follow the URL notation in Markdown to identify characters. Online editors like StackEdit allow you to double-click on a name and add an URL (with CTRL+L). By writing PER in the URL part (for the tag "PERSON"), you will then be able to use charnetto to extract the annotated entities and generate a character network.