The research project I participated in aims to customize Amazon's Alexa to provide technological support for people experiencing memory loss. The primary project owner is Libby Ferland, a PhD student at the University of Minnesota, Twin Cities.
Journal
Week 1: June 16, 2017
The majority of my first two weeks will be spent teaching at Dr. Gini's Summer Computing Academy for middle and high school girls. I really enjoy teaching and it was my driving reason behind reaching out to Dr. Gini this summer. I wasn't expecting to do research this way, so I was really surprised to get this opportunity! Of course, I'm still really glad that I am still able to teach.
I find that many students who have little to no experience in programming are afraid to jump into new material. Too often they believe that their lack of experience will cause them to fail, even before they even give themselves a chance! I try to inspire every student to be open to learning programming as a tool to pursue their other interests.
This week at camp I noticed that a student tended to give up early on the lab work and entertain herself with Buzzfeed quizzes. I showed her how to make similar quizzes using Python using if/else statements and she had a lot of fun making her own quizzes and showing her friends how to make their own.
Week 2: June 23, 2017
The camp is over now and I'm sad to see all the students go. It was really fun to get to know them! I also really enjoyed helping them with their projects -- it's really inspiring to see these girls coding small text games in Python or printing 3D sculptures they designed, especially knowing that two weeks ago many of them were apprehensive to start coding for the first time. Many of them promised to keep in touch, which is very sweet of them.
Week 3: June 30, 2017
My classes during my first two years of undergrad didn't require me to read a lot of research papers, so I'm not accustomed to reading academic text. A lot of the topics in this project -- artifical intelligence, machine learning, even developing for Alexa -- are outside the scope of my knowledge, so I need to do a lot of self studying to know enough to be useful in this project.
I found this article from AAAS that discusses several different strategies on reading research papers. I also started Andrew Ng's Machine Learning course freely avaliable on Coursera to supplement my research.
Developing a calendar/reminder application for Alexa is actually more troublesome than I originally thought. As of right now, the Alexa Skills kit does not support modifying calendar events, just reading them. We'll have to create our own framework that manages an external calendar. I have some experience using the Google Calendar API to create Python scripts to bulk import my classes to Google Calendar, so at least I have some experience to build on.
Week 4: July 7, 2017
Dr. Ng's course is really dense.
Although that just be my perspective -- all the topics of this class are very new to me; I've never studied anything related to data science at all. Despite only have finished a week of this class, the lessons introducing the concepts of machine learning and general artificial intelligence have really helped my finding academic papers. Of course, I still struggle to understand most research but knowing the difference between different AI problems (e.g., unsupervised vs. supervised learning) helps me understand the problem that is being researched in the first place.
Week 5: July 14, 2017
My research group started a Mendeley group to share research papers among us. I really like using Mendeley for keeping my research papers organized. I started collecting a bunch of papers that seem interesting to me and that I want to read again later in the future after I have more research experience.
I wish that the introductory courses at my school encouraged students to read research papers more frequently. An 19-year-old student unsure about what they want to study in the future can find something interesting from exploring research papers. More students should be aware of all the free journals and articles that they can access through their university.
Week 6: July 21, 2017
I feel like there are many misconceptions about research among undergraduates. Most of my peers view research as a job that requires immense amounts of prior knowledge and experience, which is why a lot of undergraduates are so apprehensive to get involved - they feel like they don't know enough! This really isn't the case at all -- the point of doing research is to learn something new from it, so all is well as long as you keep learning.
I'm really glad that the DREU program exists and that more undergraduates have more opportunities to get introduced to research.
Week 7: July 28, 2017
I find it incredibly frustrating that the actual language processing in the Alexa development kit is completely black-boxed. It would be immensely helpful to read documentation on how conversations and context clues were processsed, rather than plugging in some utterances and hoping that Alexa will understand your code.
From my perspective, the incredibly-simplistic UI of the Amazon Skills Kit also hides many, many potentionally useful features. Maybe I am missing something big, or this is just how developing for major frameworks tends to be done.
Week 8: August 4, 2017
Dr. Gini introduced me to an incoming freshman that was interested in starting a web development program for underprivledged high school girls. It will take place in Feburary 2018. I'm really excited since web development was how I was introduced to programming via Neopets, the crux of my virtual life as a girl in the early 2000s. As an eight-year-old, learning web development first was more motivating because you can actually see your creation alongside other websites. I really liked the feeling that I had my own little corner of the Web. The feeling of playing on Neopets is a lot different than the feeling of working on a programming assignment for university, despite the fact that they're both programming.
Of course, web development has overcome major changes since the early 2000s and my knowledge is really, really outdated (it's odd to say that as a barely third-year undergraduate) so I will have to brush up on my knowledge, which is alright.
Weeks 9 and 10: August 18, 2017
This is the last week of my DREU experience. I will combine the last two weeks in a single entry, as I went travelling to California to visit my father the previous week.
Honestly, I can say that this experience is not what I was expecting what I first started. When Dr. Gini first told me about starting research, I was really worried that I would do really, really badly because of my lack of experience. Research is literally just a learning experience, and I really enjoy the feeling of exploring new topics and methodologies. There's no supervisor telling me exactly what and how things need to be done, and I get the feeling of owning my own little part of a research project.
I think I want to go to graduate school and get more involved in academia. Doing research gives me a better sense of purpose when I take classes (like Dr. Ng's class this summer), which is really motivating. In the short term, I am going to continue this project during the school year for credit as a directed research project.
I want to start my own small research project during the school year involving natural language processing. I showed my mother the Alexa and she found it absolutely facinating, although there was one major issue -- it had trouble understanding her broken English. My mom came to the US twenty years ago, and even though it's clear that English wasn't her native language, she has little to no trouble commucating with native English speakers. Our household only speaks English!
The frustration behind this Alexa project is motivating me to try to learn more about natural language processing. Voice-activating assistants will only become more relevant as the need for assistive home technology advances at its current rate. I want to make this new technology more accessable to everybody, especially non-native English speakers.
