The Project

Idea

The main idea was inspired by this Medium Article where the author, Kate Starbird analyses tweets and news sources they site. With this information she manages to build a network about the flow of information.

So we thought, what can we find that is interesting about the tweets right after a big event, such a shooting, an attack (or maybe something not terrorism related)?

The main purpose of this project is to build a set of tools to which anyone can feed a query to a database, so that he/she can directly return insights from it. We use the data available through twitter to make sense of all the opinions and facts and alternative facts lies out there. In the process we want to try to uncover some of the hidden mechanics that bring about our own perceived reality of these events and the constructed filter bubbles.

As the tools we developed were build with reproducibility in mind, we show that we can do similar analyses quite quickly by not analysing just one event, but two. In the menu on the side you can easily navigate through the different parts of the analysis for both events easily.

In case my explanation sucks (which it probably does, here’s a picture):

Data

To handle data, we used several tools:

• MongoDB to store our data and PyMongo to handle it.
• To actually receive tweets, the GetOldTweets API that can be found here.
• NetworkX for the network analysis.
• DeepMoji for sentiment analysis and classification.
• d3js for interactive plotting in Javascript.
• Other standard python libraries (numpy, matplotlib, etc.)

Analysis

To prove the pipeline of tools, we applied them to two different events:

• The 2017 Sutherland Springs church shooting. (5th November 2017) Notebook
  ~50.000 tweets
• The 2017 Las Vegas shooting. (1st October 2017) Notebook
  ~170.000 tweets

There are many ways you can consult the analysis:

• Browse them in a more interactive, article-like way in the sections on the left of this page.
• Download the Notebooks, each one related to one of the events.

Executive Summary

Throughout our analysis, we found a lot of cool stuff, here are some of the most important findings:

Las Vegas Shooting:

• Infowars and Intellihub are within the 15 most cited sources in the database of tweets.
• The distribution of the amount of tweets with #lasvegasshooting decreases exponentially over time.
• Two communities of users were detected using the Python Louvain [1] Algorithm where at least 60% of the users cited Infowars and Intellihub.

Sutherland Springs Shooting:

• On the Sentiment Analysis, something very interesting happens with user sentiment at the 1 week mark, where the DeepMoji model helps us identify some sort of mourning, exactly one week after the event, where people start referring to the shooting in a more compassionate, less agressive way.
• From community detection, we found a community with number 6 that was very homogenic in its sentiment expressed in emojis towards certain hashtags, meaning they has either only negative or only positive emojis for most hashtags. They are big fans of #2a (second amendment) but not so much of #guncontrol. They display rather positive emotions for #maga (make america great again), #msm (mainstram media, but used by far right), #nra (the national rifle association) and #tcot (top conservative on twitter). They seem rather divided on #trump with both clapping and crying emojis. Interesting to see that the most connected users in this community are ABC News and CBS News.

Team

• Ahmet Bağlan, MSc Digital Media Engineering 🇹🇷
• Duarte O.Carmo, MSc Engineering Management 🇵🇹
• Wisse Barkhof, MSc Digital Media Engineering 🇳🇱