Cristina's Project Page

Who:  I am from Utah and I joined ACCESS so I could find out if research was a path that I wanted to pursue before I got to grad school. I enjoy practicing various instruments, doing yoga, and playing videogames whenever I have a chance.

My scientific/engineering interests:  Given that the modern world is almost entirely run by computers, I couldn’t think of a better topic to study. I am fascinated by (and a little scared of) the ever-increasing speed of tech development so I hope that my chosen field of study will allow me to be at the forefront of that development.

Academic goals:  I am a computer science major with a minor in business management. After completing my Bachelor’s degree, I plan to either pursue a Master’s in CS or to go work in the industry.

Career goals:  I am undecided on my exact career goals, but I am interested in machine learning, information security, and ethics in computing, and I have no doubt that there is a job out there that combines all three.

During natural disasters and other crises, it is critical that the public is made aware of essential resources and lifesaving information. Social media has become a prominent tool for spreading information, but not all posts are created equal. This research focused on public engagement with tweets from these official sources: News media, meteorologists, local authorities, humanitarian organizations, and politicians. The aim was to determine why politicians’ tweets, when compared to those of the other authoritative sources, garnered so much more engagement from the public during the 2017 hurricane season, specifically with hurricanes Harvey, Irma, and Maria. Millions of tweets and their associated data were collected from Twitter and various methods were applied in an effort to detect different patterns in the politicians’ tweets relative to the tweets from the other sources. During this process, tweets were sorted into topic-based clusters via the Biterm Topic Model, and the sentiment analysis algorithm VADER was applied to categorize tweets as positive, neutral, or negative. Research is ongoing and currently moving towards the use of the LIWC algorithm for more in-depth sentiment analysis and regression modeling in order to determine how the linguistic features of topic use and sentiment contributed to the higher engagement rate observed in politicians’ tweets.

Role of Linguistic Features in Public Engagement With Politicians’ Tweets During the 2017 Hurricane Season Download Role of Linguistic Features in Public Engagement With Politicians’ Tweets During the 2017 Hurricane Season