Last week, I wrote about my first ever research project and publication, and I’ll continue this topic today by talking about my next project and publication at GroupLens. Actually, quite a lot of time passed between these two projects, so first I’ll give a quick overview of what did in between the projects; I might elaborate on some of these in the future with dedicated blog posts of their own.

I finished working on the death classifier project in Spring of my Freshman year of college. That summer, I had a software internship with API Outsourcing, a local financial company in Eagan, MN, so I didn’t do any work that summer. When I came back to GroupLens in the fall, I started working on my own independent project, funded by a grant that I had won in the previous spring. The topic of this project was an extension to my work from the spring, using this death classifier to make generate further insights from the data. With hindsight, this independent project was a mistake – I was nowhere near ready to work by myself without any direct support. I didn’t have the technical skills, the research experience, or even enough free time to self-direct a research project of my own, and the I never really got off the ground.

The next spring, I admitted to my mistake, and decided to work with a PhD student again. This semester, I started working on a CaringBridge new project, led by Zach Levonian.  Zach’s wanted to take our research with this dataset in a much more qualitative direction than we did in the first project. However, we were all pretty new to qualitative analysis, so we spent the first part of the semester reading about qualitative methods, and how we could apply them to health datasets. In the end, we decided to build off a recent framework on stages of an individual’s cancer journey. We spent the rest of the semester qualitatively encoding journal posts based on this framework. At first, we used physical paper, and a _lot_ of post-it notes to go through the data, but we transitioned to an online tool (which I can’t remember the name of) to make data coding and build off these encoded posts for future, hopefully getting a paper out of it in the process!

In the summer, I left campus to do a research internship program at the Civic Data Science program at Georgia Tech. In the fall, I took a gap-semester to complete a software internship at Amazon. I came back to GroupLens in December to finish up the project before an early January paper submission deadline. I rejoined the project to find out that the focus had changed drastically over the two semesters I had been gone (no surprise). A new batch of undergrads joined the project and made contributions of their own. One of these involved building a machine learning model to classify stages of the cancer journey. Over winter break, Zach gave me a new task to build a keyword classifier to do the same classification as the ML model and compare and contrast the two. I finished up these tasks before the deadline, and we were able to incorporate my new work into the paper.

Overall, we were able to submit the paper, and it was accepted to AAAI ICWSM 2020! Though this paper has only been out for under a year, it’s already at 4 citations, which means that at least somebody out there is finding this useful! For more information on the paper as a whole, check out the AAAI page, and you can access a PDF copy of the paper at this link.

Thanks for reading everybody and see you all next week!

Categories: Blog


Leave a Reply

Avatar placeholder

Your email address will not be published.