Portfolio

Part I: Final project

ILLUSTRATION: ELENA LACEY; GETTY IMAGES

Project Brief

Summary

In brief I would like to create a data visualization emphasizing the need for political content moderation due to the prevalence of misinformation and hate speech on social media platforms. our minds have been primed to pick up extreme and oftentimes negative headlines while scrolling through social media. Social media plays an important role in helping us build communities and connecting us with the larger world, though the diversity of the content we see online has created divergent experiences for different groups using SMP. Personalized feeds designed by complex algorithms (we do not understand) have led to a behavioral change that didn’t happen overnight, leading to effects that have permanently changed the way we think and behave. These algorithms are designed to make individuals stay on respective platforms longer, so it’s no surprise that extreme content is what makes its way into our social feeds.

Some of the direct effects of poor moderation of social media, is Facebook’s involvement in Myanmar, increased polarization, YouTube’s recommendation algorithm’s effect on Bolsonaro’s election.

I would like to visualize some of the negative effects of poor content moderation, the inherent biases we have that affect our behavior while using social media, and perhaps the way the AI works behind the scenes.

Project Outline

Setup: Misinformation and hate speech is rampant on social media.

Conflict:

1.Its impact on political messaging is creating polarization. 2.Increase in offline hate crimes. 3.There is a long term behavior change. 4.Echo chambers are preventing us from getting diverse perspectives. 5.Algorithms have a part to play in curating problematic content.

While explaining the ‘conflict’ I also plan to weave in examples to further drill down on the points of conflict.

Resolution: Social Media platforms need to take better action to prevent hate speech and misinformation.

Initial sketches

Some initial sketches made to depict the potential storyboard, this may change as I find more/don’t find relevant data

Using one of the datasets I was able to create an example for the types of visuals that might be helpful. In this visual you can see that on Facebook there is a much higher proportion of headlines pertaining to politics that are being shared. Of the sources that are shared, Breitbart News articles were shared the most on Facebook. Breitnart has published a number of conspiracy theories and intentionally misleading stories.

The data

Preliminary sources that I have started to collect, these include quantitative datasets:

Date Title	How will it be Used	Source
US Politicians Political Dataset	This data will be used to show the number of politicians using social media platforms	Link Here
Social Media Posts from US Politicians	This data will be used to analyze bias, and types of messaging by politcans	Link Here
How News Appears on Social Media	The data needs to be tagged by looking at words used in headlines, the idea for this dataset is to find patterns in the ways headlines are communicated in order to see if there’s variations is headlines in traditional media vs. social media	Link Here
News Popularity in Multiple Social Media Platforms Data Set	Data to be visualized to show which news sources are shared on which platforms under different categories	Link Here

Some data that is qualitative that I can work on to make visuals for:

Article/Paper Title	How will it be Used	source
Double Standards in Social Media Content Moderation	Article has info that can be used to show algorithm bias, and existing moderation methods	Link Here
Polarized Information Ecosystems can Reorganize Social Networks	Data to illustrate polarization	Link Here
Social Media Effects	Information from this article can be used to illustrate polarization and bias	Link Here

Method and medium

The method I will be using to create the visuals will be mostly limited to Flourish and Tableau, I will be using Shorthand to compile the visuals with the text.