Peter Ball's ARIA Project:聽Social Storytelling: A Quantitative Study of Narrative Patterns in Online Communities
My project models narrative in a large corpus of news articles about protests over the past six years, with the aim of both discovering meaningful patterns in narrative structure, and determining which metadata features predict those patterns. I apply the Hidden Markov model, a classic in computer science, to the detection of patterns in the sequence of actants and their interactant relations, pulling on the work of Algirdas Greimas. Articles are transformed using natural language processing (NLP) technology into symbol sequences that encode the occurrences of protesters and police, as subjects and objects throughout the text. These articles are strung together and fed into the model, which produces a system of hidden states that encodes the 鈥渘arrative competence鈥� of the journalists, a representation of the unconscious structures that inform how these characters are invoked by their authors. This model is then closely examined to find patterns of note, which are used as classes in a predictive task that helps us understand their origin.
The ARIA program appealed to me because it gave me the opportunity to take full advantage of my undergraduate degree and indulge in my love of learning in a safe environment. The program had enough structure for me to thrive, but enough freedom to explore, and paired with fantastic supervisors, it seemed a natural choice.
My first learning objective was to expand my understanding of structuralist and narrative theory. Early this summer I asked my supervisor to recommend some articles, books and authors to immerse myself in the field, and through reading these texts I have become well versed in the language and concepts of the discipline. My second objective was to gain experience building a functioning data pipeline that combined multiple programming languages and tasks. I wrote SQL queries to collect the data, thousands of lines of Python, and made sure to open up the C code that produced my HMMs and understand how it worked. Overall, I feel I now have a more holistic data science toolkit at my disposal, which will be invaluable in my career. Finally, I wanted to improve my organizational skills when it came to research and programming. I read articles on proper research note-taking and data organization, and also greatly improved my skills with git, an important version control software used widely in the technology sector.
A big highlight was the meeting where I pitched the current version of the project to my supervisors and they loved it. We had gone through many iterations in the months leading up to the project鈥檚 start date, but with each one my supervisors had critiques and I had to return to the drawing board. When I finally got the idea right and they were thrilled with it, I felt proud of myself for having the patience to continually rework my project until the plan was perfect.
The biggest challenge I faced was decision fatigue. When you are in charge of an entire project, you are constantly making small decisions about where to focus your energies, which variables are important and which are not, how to visualize something or where to store it, when to work on this or that, etc. This can be an exhausting cycle, and can lead to a sense of paralysis when considering the possible repercussions of every small decision you make. To overcome this, I took diligent notes on everything I was doing, made every decision easy to reverse by writing flexible code, and developed useful heuristics such as a random check on 20 articles after every major data cleaning decision, to make the process more systematic. In the end, this challenge gave me great habits for my future as a researcher.
ARIA has sparked a greater passion for learning in me; I expect I will look forward to reading academic papers more now, and seek out new developments in my fields, because I feel now that I am a part of the great collective of researchers and scholars producing these works. I plan to pursue graduate school, and this summer鈥檚 experience has given me skills and deliverables that will be valuable in that pursuit. And should I end up in industry, the communication, organization, and programming skills I developed will be of great assistance.
Thank you to the Arts Undergraduate Society for providing the Arts Student Employment Fund Award that funded my summer. It is a privilege to attend a university where student organizations make high quality experiences like the ARIA program accessible to their peers.
