University of Maryland


Adversarial entities around the globe continue to spread disinformation on social media and have revealed a severe vulnerability in the security of the United States and its Western allies. The design and development of false information with specific embedded narratives is propagated to enrage, excite, and even change an individual’s behavior. Emotion is a key mechanism for hijacking a narrative and propagating it to the public.

We are a team of researchers dedicated to investigating the propagation of messages by examining how emotion affects whether someone will re-share content online. Our project started in 2019 and is to continue until the summer of 2022.

The principal focus of our study entails the collection of real-world Facebook and YouTube data from Poland and Lithuania, countries chosen for their strategic relevance to NATO and Europe. We are annotating samples of over 1000 public Facebook posts and 300 YouTube videos from each country for emotions and topic content. We are also conducting computational linguistic analyses on the greater dataset from 2015-2020 to examine sociopolitical topics and to examine cross-platform information spread.


Wrocław, Poland

Our research extends beyond the ostensible six basic emotions of anger, disgust, fear, happiness, sadness, and surprise. We are currently annotating, for both emotional content and coder reactions, 23 distinct emotions including amusement, wonder, nostalgia, relief, love, pride, empathic pain, and hate as well as a nuanced emotion dubbed kama muta   (Sanskrit for “moved by love”), which is a heart-warming emotion when viewing something endearing or cute. Researchers interested in more detail about our annotation scheme and method can contact us

cat and dog

An Example of Kama Muta

We will conduct multi-level statistical analyses of the impact of emotions, specific topics, and account and message features on social media sharing and engagement. 

Our project addresses critical gaps in research about how information spreads across social media. If successful, we can enhance our understanding of not only how emotion can affect behavior online, but what kinds of emotional content are most likely to make messages go viral, for good or ill.