Movie Galaxies is a digital humanities project that displays character interaction network in films and series. It is a visual-interactive platform that contains graphs and networks from 773 films between the years 1915 and 2012. The project’s target audience can range from film enthusiasts to writers to film professors and everyone who is interested in the character dynamics from the movie catalogue in the website. I can see that this project will mostly be academically alligned with Cinema and Media Studies due to their emphasis on films and characters in films. This project integrates data from screenplays, which transforms its text relationship into visual maps, creating a nice and user-friendly way to view characters’ relationships in the films.
Sources
TL;DR: They use each movies’ screenplay.
After strolling through the website trying to find their sources, I found that each movie has a resource page that redirectes you to their Harvard DataVerse page, which states that “we created a movie script parser and determined same-scene appereance of characters as a proxy of connectedness”.
Processes
It doesn’t specifically state which tools they used to make the graphs, but going back to their Harvard DataVerse page, the notes section states that “(each co-appeareance is measured as one degree unit per scene)”. Furthermore, the description states that “the data includes individual network graph data in Graph Exchange XML Format and descriptive statistics on measures such as clustering coefficient, degree, density, diameter, modularity, average path length, the total number of edges, and the total number of nodes”.
I also discovered something interesting in their stats page. If you scroll down to the Country headline, you will find “Geo Map visualizations require loading the “topojson” library.”
Presentation
Simply put, it displays the data in graphs.
Specifically, the characters are displayed in nodes of colored bubbles which has a thin edge (the lines) to connect it to other colored nodes (other characters).
Question
They stated in their Harvard DataVerse page that “even after multiple manual checks, the data set can still contain minor errors.” Since the project is deposited in 2018, I wonder if AI can now help speed up the process of checking those minor issues and help fix them? As well as helping speeding the process of adding new films?
