Character relationship visualization through graphs

How it works:

The original paper (French) describes in detail the technique that this software implements.

In short, it does the following:

• reads a csv file with the present of each character for each timewise unit (this usually corresponds to a page, a paragraph, a sequence, a minute, etc.)
• optionally reads up to two other csv files to describe one or two charactersitics of the characters
• creates an incidence graph based on the simultaneous presence of characters
• prunes out the characters that have little interaction / simultaneous presence
• produce a graph whose edges are weighted, based on the number of interactions had
• output such graph as a pdf file

Usage

Note: This repository is a fork of the original one and claims to be a cleaner version of it. If you are a student of M.Triclot, I recommend you use the original repository instead.

Clone this repository:

git clone https://github.com/adri326/character-graph
cd character-graph

Make sure to have R installed and run the main script:

R -f main.R --args "Cloud Atlas" 5 data/2013.cloud_atlas-adj.csv data/2013.cloud_atlas-attr.csv

The syntax for running the main script as-is is:

R -f main.R --args <name> <threshold> <presence-file> [<property-1> [<property-2>]]

• <threshold> is the minimum number of simultaneous presence that a character must have (a value of 10 is usually enough to filter out most secondary "noise" characters)
• <presence-file> is the csv file describing the presence of each character at each given step
• <property-1> is the csv file describing the primary property of each character (will be transcribed as the color of their vertice)
• <property-2> is the csv file describing the secundary property of each character (will be transcribed as the shape of their vertice)