Norah Abokhodair, PhD

The tragic terrorist attacks of November 13, 2015 in Paris sparked a massive global discussion via social media between those attacking and those defending Islam and Muslims. The March 22, 2016 terrorist attack in Brussels is now sparking a similar and heated online debate. This caused our research team from Qatar Computing Research Institute (QCRI) – Walid Magdy, Kareem Darwish and myself (Norah Abokhodair) – to ask: what is the breakdown of this sentiment?

We analyzed 8.36 million tweets beginning seven hours after the Paris attacks and continuing for another 50 hours. Searching for words and hashtags relevant to the event, we then identified about 900,000 tweets relating to Islam and Muslims. Our results showed that a majority (61%) of these tweets defended Muslims and Islam using the hashtags #MuslimsAreNotTerrorist (34,925 tweets at the time of writing) and #MuslimsStandWithParis (1,228 tweets). On the other hand, a considerable number of tweets – originating from Western countries such as the Netherlands, France and the U.S. – used language that blamed all Muslims and Islam for these attacks. These tweets ranged from asking their government to take action against Muslims to using the hashtag #KillAllMuslims (206 tweets). Check out the graph below for a per-country breakdown:

For example, one of the most followed accounts that appeared in our most retweeted attacking tweets was the U.S. presidential candidate Donald Trump’s.

Why won’t President Obama use the term Islamic Terrorism? Isn’t it now, after all of this time and so much death, about time!

— Donald J. Trump (@realDonaldTrump) N

In the tweet, Trump slams the Democratic Party and president Obama for not referring to the Daesh/ISIS attacks as Islamic Terrorism. When reviewing Trump’s timeline, one can see that this is part of a series of similar generalizations about Muslim culture. But can following and reading @realDonaldTrump mean you are also more likely to blame all Islam and Muslims for the tragic actions of a few?

To investigate this, our follow-up study (recently accepted to the ACM WebScience conference in collaboration with researchers from Melbourne University in Australia, will link in later post) focused on U.S. Twitter data to examine the effects of social network interactions on future attitudes. By studying pre-event interactions – including who you follow, who follows you, and the content and people you tweet and retweet – we can predict your attitude and your political stance. We confirmed that by assessing users’ stated stance based on their post-event tweets. Most impressively, we found that pre-event interactions predict someone’s attitudes towards Muslims with high accuracy, even if they never discussed the topic before. Social media messages and networks therefore have profound influence on political attitudes and shape national and international policy. Future research will allow for more accurate predictions of community response to crises and the drivers of policy change.