I analyzed sentiment in movie comments for all four Matrix films, showing a decline from positive emotions in the original to increasingly negative sentiment in sequels. The visualization unexpectedly resembles Russell’s Circumplex Model of Emotions, with dimensions for valence and arousal.
In 2024 I took part in the first week of the 30DayChartChallenge. For day two I used a scraper that I alread had prepared to download comments from Letterboxed. Letterboxd is a social platform where users can log films they’ve watched, write reviews, create and share lists, build watchlists, and follow others for movie discovery.
Since the topic of the challenge was “neo” I wanted to do something related to one of my favourite movies/franchises - Matrix :D
Data & Method
I scraped all comments from all 4 Matrix Movies from Letterboxed, which resulted in 16.221 observations. Then I used the syuzhet library to access the nrc dictionary for the sentiment-analysis of the comments 1. By this time I had never used geometric data analysis before, still with the help of a few tutorials it worked out nicely.
Interpretation
As you can see in the plot Matrix 1 is associated with anticipation, joy and surprise. The second movie Matrix 2 is associated with surprise, and trust while the Sentiments regarding the third movie are fear and anger. The last of the four movies, Matrix 4, is associated with sadness and disgust. In my opinion this plot captures the common collective opinion towards each of the Matrix Movies.
What was a great surpurise was that the axis of the plot perfectly resemble Russels Circumplex Model of Emotions, which creates a space of emotions through the main distinctions of positive/negative arousal and positive/negative valence. Dimension 1 of my Matrix plot represents valence, while dimension 2 represents the arousal/energy behind the valence (sadness is less energetic than anger).
Footnotes
Footnotes
-
The NRC Sentiment Dictionaire https://arxiv.org/pdf/1308.6297 ↩