Overview

A hackathon project initiated at MIND 2019 at Dartmouth College.

High-level idea: create narrative trajectories (kind of like these) for movie scripts mined from IMDB. Then relate various aspects of those trajectories and/or the contents of the scripts to interesting things like:

• movie ratings and reviews
• genre
• box office success (money made, number of tickets sold, etc.)
• etc.

Setup

To initialize the cluster-tools module and switch to the eventseg branch (needed to run the analyses on the Discovery cluster), run the following (in Terminal, from within the narrative_complexity directory):

git submodule update --init
cd code/cluster-tools-dartmouth
git checkout eventseg

Install all project requirements by running (from within the project folder):

pip install -r requirements.txt

Then run (from within Jupyter Lab or a Jupypter notebook) any of the .ipynb files.

Presentation

Our hackathon presentation may be found here.

Brainstorm

Potential directions for the project

• Create a database of semantic content of movies that people might want to refer to in order to select an ideal stimulus space for functional alignment
• Analyze ISC across naturalistic datasets to assess whether narrative complexity is related to similarities in functional topography
• Use narrative complexity to predict ratings and gross value of films
• Assess consistency of the narrative across different segments of movies
• Compare consistency of movie content and critical reviews