Researcher of the Month
June 2006
Matthew Marge
Majors: Computer Science, Applied Mathematics & Statistics; Honors College, Class
of '06
Research Mentors: Dr. Amanda Stent, Computer Science; Drs. Susan Brennan, Richard Gerrig, and Greg Zelinksy, Psychology
"I hit the jackpot at Stony Brook," Matthew Marge reflected, hours before going to
an end of the year ceremony sponsored by the Undergraduate Academic Affairs (UAA)
office where he and fellow Stony Brook students who had received prestigious national
scholarships this year were being honored. Matthew was one of two students nationwide
selected for the St. Andrew's Society of the State of New York Scholarship, a $15,000
award funding study in Scotland, and enabling Matthew to pursue a one-year MSc in
Artificial Intelligence at the University of Edinburgh beginning this fall. For Summer
2006, Matthew is working at the Naval Research Laboratory in Washington, D.C.—a great
summer internship brought about through an initial contact made at a Human Robot Interaction
conference he attended. Recently, Matt also attended the 2005 CICLing Conference [on
Intelligent Text Processing and Computational Linguistics] in Mexico City and presented
"Evaluation of Text Generation: Automatic Evaluation vs. Variation."
The many exciting opportunities that have come Matthew's way — including being awarded
the Barry Goldwater Scholarship in 2005, having wonderful summer research opportunities/internships, co-authoring
publications, and giving presentations at national professional conferences — all
stem from an initial foray into research duing his sophomore year; and from following
up on his interests and passions. What ended up being the subject of Matt's 
Honors College senior thesis, for example, developed out of exploring his research
interests in psychology and language during his sophomore year; as a result, Matthew
had the opportunity to do interdisciplinary faculty mentored research with Professors Amanda
Stent (Computer Science); Susan Brennan (Psychology), and Richard Gerrig (Psychology)
on an automated course evaluation system project (entitled the Stony Brook Rate-A-Course
System). In addition, Matt also conducted research on eyetrackers with Prof. Greg
Zelinksky (Psychology). Through his mentors, Matt was also introduced to valuable
off-campus contacts in his field as well.
In addition, Matthew's all out involvement with the Robot Design Team where he became
immersed in robot competition/design and eventually served as co-leader of the Team
under the guidance of Professors Amanda Stent, Dimitris Samaras (Computer Science); Imin
Kao and Peisen Huang (Mechanical Engineering), initially stemmed from following up
on a curiosity (What is the robot team all about?), and attending the team's last spring semester meeting late in his sophomore year.
Through the Robot Team, Matthew not only became friends with great faculty and student
colleagues both on and off campus (at the AAAI
conferences), but also came to know "Jester" and "Navbot" (the stand-up comic robot
that won Audience Favorite, and the navigation search and rescue robot that won the
Technical award, respectively, in the 2004 and 2005 American Association of Artificial
Intelligence Robot Competitions), as well as "Pearl" (a robot chef he worked with
during the summer of 2004 at Carnegie Mellons's Human Interaction Institute while
working under the mentorship of Prof. Sara Kiesler), and learning, among other things,
the proper tools needed to make crème bruleé!
Asked how research has enhanced his education, Matthew replied:"Research has been my education. Really..I would never learn what I learned in the classroom. Especially in my field, you don't learn this kind of stuff as an undergrad. You don't learn anything about HRI (human-robot interaction), about the applied aspects of human computer interaction, & designing dialogue systems…you don't get to do that kind of thing without research. You might not find your passion taking classes. That's why you should get involved in research. You can extend your passions from classroom with research. "
Matt expanded on this theme, stating how research "geared the way [he] chose courses. Interface design, robotics. Those kind of choices...The kind of specialism I'm going for in Edinburgh in AI is with a specialism in natural language and language engineering. My research has shaped how I will select my graduate courses. I know what I want to do. I know what my plans are." Yet with all these pursuits, as well as the well-spent time serving as a Stony Brook Student Ambassador, Matthew has also managed to fit in other passions as well--including badminton, squash and bowling! Below are some excerpts shared from his interview wKaren Kernan, URECA Director .
Karen: What kind of research do you do? How did you first get involved?
Matt:My area of research is human computer interaction. I've been working in this field
for about 3 years now. And I first got involved in my sophomore year. I can tell a
little story. ..There was an honors college open house. And I went to it. Prof. Richard
Gerrig had emailed the whole list of Honors College students asking for volunteers
to come to the Honors College open house to answer questions for new students. And
I went and met up with him. And we talked a little bit. I talked about how I was interested
in getting involved in research. After hearing about my interests, he said, "I know
this professor in computer science that you might want to meet." So I got to meet
Amanda for the first time through Professor Gerrig. . . that's how I got involved
with the research. . . . After that [first project that I became involved with on
natural language processing ], Prof. Stent also asked me about other interests. I
talked about how I'm interested in speech. We looked into working on a spoken dialogue
system project that would actually be a spin-off of a one of the special topics classes
for seniors. Speech applications. I got involved with building the Stony Brook Rate-A-Course
System. And that became my honors thesis.
So you've done interdisciplinary faculty-mentored research in Psychology & Computer
Science for a few years, then?
From the summer before my junior year, I worked on that system under an REU (NSF).
Then I remained working on it during the semester. Professor Susan Brennan (in Psychology)
became my third advisor (in addition to Profs. Gerrig and Stent). Then, I worked on
the project that summer and fall. I met with Prof. Brennan and asked her about opportunities
during the summer. She had conducted research at the Human-Computer Interaction Institute at Carnegie Mellon University. I asked her about possible internships…She gave me
some contacts. And I got to contact Sara Kiesler at Carnegie Mellon. She invited me to her lab for that next summer, before my senior
year. Last year I worked over there working on human robot interaction research.That's
how I got interested in human robot interaction.
Was it also in your sophomore year that you first became involved with the robot team?
Actually in May I met with the robot team-- the last meeting of the year!--as a sophomore.
I wanted to see what was going on with the robot team, I wanted to see what was going
on with robotics. I didn't know anything..and I was very curious I wanted to see what
it was like…Every year, they've been at a international competition since 2003. Usually
the AAAI competition. That year, we went to San Jose in California. It was an amazing
experience. . . We used part of Diana David's senior project-to build a robotic interface..We
built around that a whole robotic hull. We extended the conversation system of the
robot so that it could tell jokes and trivia. We then named the robot JESTER, the
Joke Entertainment System with Trivia Emitted Responses.
I remember JESTER! Your team presented this at URECA's Celebration.
Yes, last year at URECA's Celebration. And at the AAAI 2004 Robot Competition. We only had 2 months to work on that project. Very tight scheduling. So when we
went to the competition, we ended up winning the audience favorite award for the competition
where success really depended entirely on how the audience perceived the robot. The
name of the competition was the open interaction competition where the robot navigated
the conference floor and interacted with people, like a robotic host . . . we had
a compartment in the robot that even had cookies! There were a lot of kids at the
conference which was great. A lot of kids came up to the robot to interact with it.
Then they came back later with more friends and it was exciting to see that. People
had a lot of fun interacting with JESTER at the competition. And judges took notice
of that!
How much of a link is there between the work you did with JESTER for the AAAI Competition,
and the work you did on human robot interaction when you were working at Carnegie
Mellon's Human Interaction Institute?
I learned about how to develop an intelligent conversation system which is something
I'm really interested in.. I used the same kind of work with Pearl the next summer
[at Carnegie Mellon]. We used something called artificial intelligence markup language
(AIML) which is a language that allows you to fetch responses based on key words that
are inputed to a computer. That summer with Jester, I learned how to do that. And
at Carnegie Mellon I used the same technology to develop the conversation system for
Pearl, fetching responses for different tools. The people at HCII were already using
it. And I was able to use my skills in the area. . .
That summer (2005) at Carnegie Mellon was great too. We worked with a robot called Pearl which is a humanoid robot designed for analyzing human robot interaction. We were looking at how a robot assesses the expertise of a person in a specific field. In our case, Pearl was a robot chef. We wanted to see how people interacted with her in cooking experiments. So we gave subjects a task of finding the right tool on a webpage. There were 5 different tools presented to people. They had to select the right tool. For example, they had to find a ramekin. They had to select the right picture. They could ask questions to Pearl about details about the ramekin. We varied the amount of detail that Pearl initially gave. Pearl could say something like: "First you need a ramekin. Find a ramekin. " Or: "First you need a ramekin. Find a ramekin. It's a round shape... " and it gave a lot of detail. We wanted to see how people reacted based on those detail modifications. We recruited only experts and novices in cooking, which we got through a pre-test. We found that experts preferred Pearl when it gave no detail (because they were being timed); and novices found Pearl patronizing when it gave them no detail (i.e. the robot is assuming that I know what a ramekin is). These kind of details for expertise are very important in human robot interactions.
Was the research/design experience you did with the Robot Design Team very different
than your experiences doing independent faculty-mentored research?
What was really great was that Amanda Stent was my advisor for both the robot team
and for my research. It just felt natural. I kept a good relation with all my advisors.
It's really important to try to develop a friendship with your advisor. That's what
I hope to have achieved. A mutual friendship seems to rub off. You get more freedom,
and it works out reciprocally. We share what our interests are…it worked out very
well.
Also, I learned what it was like to be a team member of a robot team. …The robot team is very demanding. The toughest thing is to really make sure the team gets organized and has a common goal. That was my role that summer, I was the co-leader. We had a very strong schedule. And we held up to the schedule. It worked out really well. Everyone had a common interest. Everyone was so aware that we only had two months. The team turned out to be fantastic…
That [first] year we only had electrical engineers and computer scientists. That following year ..we wanted to make a goal of recruiting mechanical engineers so we could develop and build robots, in addition to using laptops and such. So the next year, we built a robot for navigation called the NAVBOT -which was a search and rescue robot. We had a team of 8 or so Mechanical engineers who built the hull…we had electrical engineers work on circuit designs and so on…and we had computer scientists do some of the programming for navigating the robot. The robot team involves a lot of different fields of robotics. With Jester it was human robot interaction and conversation. With NAVBOT it was more the navigation and planning…
I understand you've been to several conferences, with the Robot Team as well as through
your independent research. Was this a positive experience for you?
The first one, in July of 2004, that was the AAAI competition for JESTER. That summer,
we found out…Amanda and I got a paper published for the work we did on text processing.
Amanda asked me to present the work at the international conference in February. That
February of 2005, I went to Mexico City and presented my work on evaluation metrics
(or text analysis programs). And it was motivating....absolutely!. . . I got to bring
my parents along. . . It was an amazing trip… I got to meet researchers that shared
my research interests...The HRI conference is where I got to meet a lot of leaders
in the very young and emerging field of human-robot interaction. I got to meet a lot
of people that I would not get to meet anywhere else. I got to meet people like Cynthia
Breazeal at MIT. She worked on the Kismet robot. You might have known about that robot.
It can show emotion. It's very interesting stuff. I got to meet up with the people
that I had worked with at Carnegie Mellon. I got to meet people from all over the
world. . . it's a worldwide experience.
You were awarded a URECA Travel Grant this past spring to present at the conference
in Salt Lake City. What do you think you learn by presenting, and attending meetings
such as these?
Every experience you have at presenting your work makes you a better presenter. That's
the kind of thing you rarely get taught. And that's the kind of thing you need to
do to become an effective researcher. Presenting at conferences is critical.
Also the URECA Celebration [where presented a poster] was a valuable experience too. The other conferences were structured around formal presentations. At URECA's research poster event . . . you are able to expose your research in a less formal manner, and speak with people not always in your field of expertise. You get to speak to them on an appropriate level where you build the same playing field, and you are understanding things on the same level. You build that skill of being able to speak about your research to anyone. That is important. If I'm a database expert I still need to speak to those with less expertise. You need to be able to design your presentations for everyone. That kind of skill you learn from poster sessions.
So soon you'll be off to Scotland?
In the fall my plan is to go to the University of Edinburgh and pursue a masters in
artificial intelligence [with the St. Andrews Scholarship of the State of NY]. . .
I'm going actually to get to work with one of the leaders in the field of spoken dialogue
systems. I also get to meet with the leaders in the field of robotic planning. . My
primary advisor at Edinburgh, her name is Johanna Moore. She used to be at the University
of Pittsburgh. My advisors here, Amanda Stent and Susan Brennan, have worked with
her very closely. And everyone knows each other very well; it is a super small world.
I got to work with people that my advisors already know. So I can stay in the network.
I know that applying for national scholarships (such as the St. Andrews Scholarship,
or the Barry Goldwater Scholarship) can be time-consuming. Was there someone who helped
you with this process?
Absolutely! Rosemary Effiom is a fantastic person to get to know. I don't know what
to say. . . there are so many good things to say about that whole office and about
Rosemary. Rosemary is the person to meet if you want to go abroad and are looking
into external scholarships. Did you know she won the campus life award this year for
best advisor?
Do you have any advice to offer to other students?
This is general…but it really is so important to get involved. Find anything that
you're interested in. You're in your major for a reason. Find out why you are in this
major. Look into what kind of research is beng done. And look into getting involved.
You can go and visit any professor. Email them. Read through their publications. That
's the first step. After that first step, you'll be exposed to so many positive things…..positive
experiences to get through that door…overall, I suggest you get involved with campus
life, being involved with what your passions are. . . Also, I want to send a message
saying that Stony Brook can take you anywhere you want to be. It really is about how
much you put into it. Never feel held back. You can pursue your dreams at Stony Brook.