Main file is network_drawing.py for main code, written in python 3.

NOTE: the scripts are provided as is and without warranty

Libraries needed:

matplotlib
planarity
networkx (to go with planarity)
nltk


Procedure:

1. Input the name of text file (story). By default, Python assumes encoding of text file to be utf-8. If not, please amend it on line 9. 

It will take a few minutes to generate list of characters, depending on length of text file. 

2. When done, combine the characters which refer to the same character, manually or automatically.

Input format:
<list of indexes that refer to the character>
<Desired name of character>
...
end

A sample input is as follows:(Or you could copy the input from input.txt)
NOTE: Input will vary for different stories.

0 8 11 14 31
Hercule Poirot
1 20 23
Mr. Jack Renauld
5 7
Mrs. Renauld
12 28
Marthe Daubreuil
4 46
Madame Daubreuil
17 39
Jeanne Beroldy
3
M. Hautet
end

3. Afterwards, try to find a sharp decrease between 2 characters in the graph that will show, and count the number of characters before the drop, which is used by the program. This will be the number of main characters. Note that the characters are 0-indexed. The graph will be generated using matplotlib, with the x axis representing each character, while the y axis represents the semi log-y frequency of each character appearing in the story. For more details, refer to our research paper. 

4. When the programme is done, the output file will be on "network_drawing_output.gexf" and "network_drawing_output_pmfg.gexf"(after running the planar maximally filtered graph (PMFG) algorithm) in the same directory. Open this file with gephi software to visualise the network. 

5. Adjust the nodes (characters) to observe the networks shown in our report. Note that the settings in Gephi should allow the names of the characters and the colour of the edges to be visible.