Design Process

Initially we were not clear on what we wanted to visualize in the scope of this design session. We had a relatively large dataset that needed to be restructured into a single table with observations and features (or key/value documents). Visualizing summarizations and features of the data, as we saw in the design lectures, did not seem to capture our research purpose . But after receiving feedback from the teaching assistants we focused more on visualizing the network aspect of our data. In the end our design approach was focused more on showing a network as graph with nodes and edges and using these components to include the other features of our dataset.

The miro board

Diverge Phase

Some of the interesting designs from this phase include:

This was one of the first designs in the diverge phase. Even tough it does not focus on the network aspect of the data, it can help visualization certain features like sentiment scores of tweets as the population of tweets evolve from day to day

This is again one of the earlier plots that show a crude network that tries to introduce a notion of centrality with a directed graph possibly showing the central nodes (tweets/accounts) in a sentiment group.

Emerge Phase

Here, the initial stage was to cluster designs together. However, the bulk of the processing focused on merging designs, reworking previous network graphs and comparing different clusterings of nodes in a network.

A network as in the first picture might be too convoluted to show the most connected nodes in a group (or across a group). An arch digram, of which this is a very crude representation, will allow us to show the density of links around a node more effectively.

DELANI~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Diverge Phase

(a) Among a list of popular anti-vaccine tweets that were withheld (tweets with large number of favorites or retweets), a night map of the world can be shown with the lights on in every country except the countries that have withheld the tweet. (b) Among all anti-vaccine tweets that were withheld, a distorted map of the world can be shown to illustrate which countries are flagging disinformation. (c) A pictograph visualizing the device used to produce the tweets. Where each icon represents a certain number of tweets (e.g., every Apple logo shown represents 100,000 tweets).

Emerge Phase

To cluster the ideas above together, one can use the globe of the Earth to visualize the evolution of popular anti-vaccine tweets that are withheld by country over time.

• Organize the withheld tweets in chronological order by day.
• Each 360 turn of the globe will represent the passing of a day.
• To move between adjacent days that have produced withheld tweets, click on the globe, it will spin X times representing the X days between those published tweets
• For each day, countries that withheld tweets will be highlighted
• When you zoom in on the country, you’ll see a stacked pictograph of the number of withheld tweets for that day with the pictures illustrating the source of the tweets.

The Disinformation Dozen

In March 2021, the Center for Countering Digital Hate (CCDH) published a social media analysis report on a sample of 812,000 anti-vaccine content from Facebook and Twitter between 1 February and 16 March 2021 to investigate the dispersion of disinformation regarding the COVID-19 vaccine. The report highlights 12 individuals (listed below) that are at the head of spreading disinformation on these platforms by using their follower count and their anti-vaccine content volume and dubs them the “Disinformation Dozen” (DD). Of the 120,000 anti-vaccine tweets analyzed, 17% was associated with the disinformation dozen. For Facebook, 73% of the 689,000 posts were associated with this group.

The Disinformation Dozen

An analysis of the Lopez Twitter data set was conducted to see how much of an influence the DD has had on the data set.

Diverge Phase

The faces of the disinformation dozen will be used like a word cloud, where the size of each person’s face is proportional to the percent of tweets in the Lopez data set that originates from them.

Emerge Phase

A network of the DD will be created to see if and how their disinformation spreads among themselves. The faces of the DD will be used as nodes and the edges will represent the number of retweets/favorites that are shared among one another.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~