blog

printing code: final proposal

for my thesis project, i've been working on extracting sentiment from hate speech on the internet with python and assigning it to a color spectrum in processing.  the results are then printed on fabric and used to create high-fashion, couture garments.

i mention this because i've had a related idea kicking around a while for my final for this class that i'm really excited about and think might work.  and, now that i know the basics as far as the language processing is concerned, i think i might be able to translate it effectively so that it happens entirely inside processing without the need for running an external python script.

i'd like to do a series of prints that visually examine gender bias in articles about rape that have been published in the mainstream media in 2012-2013.  there have been a ton of high-profile rape cases over the past year that have garnered the attention of the american (and international) public, such as steubenville, torrington, the delhi bus rape, rehtaeh parsons, and a series of UNC rape victims who have been threatened with expulsion for speaking out about their assaults.  each time the mainstream media covers one of these cases, there is a huge outcry from the blogosphere about how they are biased towards the accused rather than the victim, perpetuating the state of rape culture in the world and making it even harder for future victims to seek help.  i want to examine this claim in a visual manner that could potentially have a greater impact than yet another blog post decrying the state of media bias today.

i've pulled several articles so far about various cases, which i intend to use as the corpora for this project.  i'd like to create outline paths of a man and a woman as SVGs, which i would then load into processing.  from there, i would plot points along the paths that correspond to each word in the article examined (the man and woman would have the same number of points).  if a word is determined to be biased towards the man, a line would be drawn to the corresponding point on the male path; and the same would happen with the female version.  i would also adjust color depending on how many biased words there are towards each gender, making the color darker and richer as more and more words accumulate.  the male and female paths would be laid on the same canvas -- the idea is that the more biased towards the man the article is, the clearer the outline of the man would be, and vice versa.  i'd do a separate print for each article i examine.

here are some sketches i did in photoshop, that kind of illustrate what i'm going for:

example-withoutline.png
example-nooutline.png

i also built a simple text parser in processing, using an article about rehtaeh parsons as a sample, to test translating the text to geomerative.  here are some results for various efforts:

Screen Shot 2013-04-13 at 5.28.02 PM.png
Screen Shot 2013-04-13 at 6.35.03 PM.png

the major thing i will have to focus on immediately is accurate language processing, which will also entail a ton of research over the next week; but since i've already dealt with this to some extent for my thesis, i'm pretty sure i can make progress on this relatively quickly at this point.  i also need to find recognizable male and female forms to trace the paths of, as well as figure out size and placement so that the lines will be enough to enable identification of the forms once there are enough of them present.

Roopa Vasudevan