Project #10: Translation Device
Description
In this project, we learned about data and APIs, and we are aiming to create a translation device in p5.js that translates external JSON datasets into feminist data. While reading Data Feminism by Catherine D’Ignazio and Lauren F.Klein, Chapter 4: What Gets Counted Counts, appeared to be most inspiring to me. what we see and what remains invisible are equally important for data researchers, yet the invisibles can be easily overlooked and results in data bias. The data being collected might not precisely indicate the reality, but rather results of collective oppression.
Technical components that I explored in this project:
- API and JSON
- Dot notation
Design Process
Movie industry quickly popped into my head as soon as I decided to do something related to Principle 4 of Data Feminism. What really inspired me was this graph(below), showing the 10 year rolling average of return on investment(ROI) for movies that have production budget data listed on the website boxofficemojo.com:
At first glance, it might seem like the movie industry was extremely profitable from the 1930s to 1970s, but started to take a downward turn in the late 70s. However, what’s tricky about this dataset is that boxofficemojo.com only came into existence in 1999, and each film that is included in this graph released before 1999 are only added if people still remember about the film in 1999, which are more likely to be great and profitable films. Which is a great example showing how counting or classification systems can be flawed. In this case, both quantitative and qualitative thinking are needed for us to deduct this result, and simply looking at the graph can be deceiving and result in false conclusions.
Therefore, I tried to find a movie-related API, and finally settled on The Movie DataBase(TMDB) , which is a pretty authoritative website for movies, TV shows and viewer reviews. I followed the API instructions and applied for an API key for this project. After studying the datasets they provide, I decided to work on the average vote score and the vote counts of movies. For most people, the score of a movie can greatly influence viewers' decisions, yet the amount of votes are the hidden component deciding whether the score is accurate or not. Therefore, before judging a movie by it’s score displayed, it would be best to look at both datasets to make the best decision.
Reflection
I truly enjoyed the process of working with APIs, partly because it gives me the feeling of sneak peeking into some of the information that I once thought was classified, partly because it is so useful in many aspects. With a few lines of code, our little programs are now connected to some of the most powerful databases in the world, and this idea just sounds unbelievable to me at the beginning. It wasn’t easy to find and understand each APIs, and it required some learning in order to use the information. I truly hope more organizations could be kind enough to be providing well-documented APIs for people to access in the future, it really provided valuable insights to me in many ways.