Sentiment Analysis of Final Fantasy VII and Final Fantasy VII Remake (with Comparison)

A quick note: I was working on this project while Final Fantasy VII celabrating its 25th anniversary (June 16th, 2022). That was a nice little coincidence. 😊

Introduction

First of all, I would like to clarify what Final Fantasy is for those who haven't heard it before. Final Fantasy VII is a JRPG (Japanese role playing video game) developed by Square Enix (formerly Squaresoft) released in 1997 for the PlayStation console. It is the seventh main installment in the Final Fantasy series. The game's story follows Cloud Strife, a mercenary who joins an eco-terrorist organization to stop a world-controlling megacorporation from using the planet's life essence as an energy source [1]. This game got remade in 2020 with new gameplay and latest graphics.

I wanted to make a Natural Language Processing (NLP) related project and while I was researching Sentiment Analysis piqued my interest. So I decided to make the project on this topic. Afterwards, I came across this great paper titled Analyzing Movie Scripts as Unstructured Text [2]. This paper is the main inspiration for my project.

Then I found the dialogue scripts of Final Fantasy series perfectly compiled on Kaggle by Tyler Huxtable and start working on the project [3]. Here's a link to the dataset.

Project Goals

• Measure the sentimental intensity of the dialogues in these two games. With the help of NLTK (Natural Language Toolkit) we can measure statements' positivity, negativity and neutrality scores

• Inspect the course of storyline in terms of positivity and negativity of the dialogues

• Compare the positivity of the narration between the original game (1997) and the remake (2020)

Sentiment Analysis with Vader Module from NLTK

I've used nltk.sentiment.vader [4] to measure the games' lines' positivity, negativity and neutrality scores. You can find the source code of the module here [5].

There are about 6000 lines of dialogue in FF7 and 8000 in FF7 Remake. I got the sentimental intensity of these lines and smooth the result a bit to see the trend more easily. Here's a graph comparing the two games.

These two graphs show the dialogue distribution of the characters.

More details are on the Jupyter Notebook... 🏃‍♂️

References

1. "Final Fantasy VII - Wikipedia". https://en.wikipedia.org/wiki/Final_Fantasy_VII

2. Lee, Seong-Ho & yu, hye yeon & Cheong, Yun-Gyung.(2017). Analyzing Movie Scripts as Unstructured Text. 249-254. 10.1109/BigDataService.2017.43.

3. "Final Fantasy Dialogue Scripts | Kaggle". https://www.kaggle.com/datasets/tylerhuxtable/final-fantasy-dialogue-scripts

4. Hutto, C.J. & Gilbert, E.E. (2014). VADER: A Parsimonious Rule-based Model for Sentiment Analysis of Social Media Text. Eighth International Conference on Weblogs and Social Media (ICWSM-14). Ann Arbor, MI, June 2014.

5. "NLTK :: nltk.sentiment.vader". https://www.nltk.org/_modules/nltk/sentiment/vader.html