The Machine Learning Center at Georgia Tech (ML@GT) is home to many talented students from across campus, representing all six of Georgia Tech’s colleges and the Georgia Tech Research Institute (GTRI).
These students have diverse backgrounds and a wide variety of interests both inside and outside of the classroom. Today, we’d like you to meet Ian Stewart, a fifth-year Ph.D. student in human-centered computing. When Stewart is not analyzing how language and computers come together through natural language processing (NLP), he can often be found running, baking (with lots of butter!), or finding a new aesthetic to try in his wardrobe.
Name: Ian Stewart
Hometown: Ipswich, Mass.
Advisors: Jacob Eisenstein and Diyi Yang
Current degree: 5th year Human-Centered Computing Ph.D.
Other degrees: B.A. in Linguistics from Dartmouth College
Tell us about your research interests:
The way that people talk online can reveal a lot about their social expectations. What makes new words stick in a community? How does a person change their speech style to match their assumptions about their audience? When does a person’s attitude toward a topic influence the way they speak, even when they aren’t discussing the topic? Studying these questions in online discussions gives us a way to test social theories at scale, which can help us design better discussion platforms and allow the next generation of natural language processing tools to capture these social cues. I use NLP and statistical testing to test sociolinguistic theory at scale across a variety of discussion contexts and platforms.
What drew you to wanting to research these areas?
I studied linguistics as an undergraduate and I always found the “social” to be the most interesting part of language. For instance, what makes people perceive one kind of speech as a “dialect” and another kind of speech as “normal”? When applying to grad school, I wanted to combine my interests in sociolinguistics with my interests in quantitative modeling, and Georgia Tech was a good fit.
What motivated or inspired you to pursue a Ph.D.?
I was lucky enough to work in linguistics research during my undergraduate degree and I realized that research was a good fit for my interests. Doing a Ph.D. gave me the freedom to pursue complicated questions and to develop my computational skills.
What is it about machine learning and natural language processing that excites you?
NLP is a tool that can help researchers tackle difficult questions at scale. How can I find all posts written in Spanish? What if I needed to extract all dependent clauses from a sentence? Using an NLP model like a parser or a tagger lets me find rare linguistic patterns in a way that is very difficult to do manually.
Earning your Ph.D. requires a lot of time and hard work. What has been challenging, rewarding, or unexpected about this process?
The most unexpected part of the Ph.D. has been being exposed to very different perspectives on research. My research floats between machine learning and social science, and absorbing the variety of “in-between” perspectives has really warped my brain. For instance, how do we make ML systems that are not just accurate but also fair? It’s not always fun, but it has forced me to be open-minded.
What is your favorite place to hang out in Atlanta and why?
I have a soft spot for Amelie’s Cafe in West Midtown. It’s fancy and cozy and punk all at once somehow. That’s a very hard aesthetic to pull off. Also, the seasonal drinks are unreasonably good.
You have a great sense of style. Do you have any fashion icons or inspirations? How do you go about putting your outfits together?
I love bright colors, and I like testing out different color combinations. Sometimes I’ll borrow an aesthetic from TV shows or movies. I recently finished “The Good Place” which has a lot of bright outfits and patterns that I really liked. People should always experiment and play with the way they present themselves, no matter how old you are.
Tell us about some of your hobbies:
I’m running the Publix Atlanta Marathon this year, so I’ve been running about 20 miles a week to train for that. I also play several musical instruments, mostly the guitar nowadays. It’s very cathartic to do something manual after working on a computer all day.
What is your favorite Georgia Tech experience and why?
I love visiting the GT farmer’s market every Wednesday in the fall. I used to swap recipes with one of the bakers.
What are your plans for after graduation?
I’d like to work as a post-doctoral fellow to broaden my experience with computational social science, before starting an industry or a government researcher position.
Who is someone that inspires you and why?
Marie Curie. She was resilient in the face of social pressure against women, and she set an example for how to focus on scientific questions that matter for society at large. She was truly a remarkable human.
What is your proudest accomplishment?
My first solo-authored paper in undergrad was pretty great.
We’re now in a new decade. What are you most looking forward to in the next ten years?
I’m looking forward to a slower and healthier online social experience with social platforms that encourage critical thinking and empathy instead of knee-jerk reactions and polarization.
What is a guilty pleasure of yours?
Baking. I bake cookies and cakes with a lot of butter.
Podcast, book, movie, or TV show?
Books, nonfiction and sci-fi especially. I love anything that Mary Roach writes. She could write a book about paint drying and I would give it five stars on Goodreads.
If you could time travel, what period would you go to and why?
2100s. I would not do well in most other decades of the past…too much manual labor.
Why do you think embodying Georgia Tech’s motto of “progress and service” is important, especially in regards to machine learning and AI?
If we don’t build technology that serves society as a whole, and if we design systems that only benefit a select few, then we don’t deserve to be building technology at all. Whatever “ML” may be, it will be the backbone of most areas of tech within the next ten years, and we need to make sure that no one is left out of the equation.
What advice would you give to someone looking to pursue their Ph.D. or wanting to learn more about NLP?
Take chances, make mistakes! Start with something simple like text classification (is this tweet sarcastic?) and go from there. There is a lot of open-source tech available for students to test, break and remix. We have only scraped the tip of the iceberg in terms of potential applications of NLP, and I look forward to seeing the creative applications of language technology that people hack together in the future.