To label our text-formed posts, we will use https://www.zooniverse.org/ platform to offer a survey workflow. ‘Zooniverse’ is a popular platform for crowd-sourced image classification tasks in various scientific research.
This will be a non-profit task which relies on the public’s attention to Eating Disorders and willingness to voluntarily answer a series of related questions. A quick look of our classification interface:
To classify text data on this platform, I used the Image, ImageDraw, and ImageFont modules from PIL package in python.
By reformatting our post — manually insert a line ending every 11 word, and setting parameters for the image — text pixel position in an image, the background color in RGB, text font and size, and margin width of text box, the result of our ‘text-content’ images looks like this:
The code and results can be found and downloaded here:
