WPI Worcester Polytechnic Institute

Computer Science Department

AIRG Topics - Fall 2012

Our group meets on Thursdays at 11:00 a.m., FL 246, Beckett Conf. Room.

Dates and topics for this semester are as follows:

Aug 23
AIRG Organizational Meeting (Coordinator: Joe Beck)

Aug 30
AI Faculty
"Research Overview"

Sept 6
Dmitry Berenson
topic tba

Sept 13
Joe Beck
"Graduate Student Survival Skills"

Sept 20
Dongqing Xiao
topic tba
    Machine translation evaluation plays a great role as "Moses" in the development of machine translation techniques. However, the performance of state-of-art string similarity-based MT evaluation metric is invariably considered to be inadequate. Although, researchers have successfully applied linguistic features to reinforce it, the gain of performance is limited. As MT continues to develop, there are various online machine translation services which have become the routine use for web users. This process is producing user feedback on a large scale. User feedback has often been proposed as a solution for improving machine translation systems. The basic motivation is that users will be able to make disambiguating choices, take post-edit actions which fall outside of the machine translation system's capabilities, or supply extra-linguistic knowledge necessary for language analysis or language generation. Until now, relative researches on utilizing expert feedback in MT automatic evaluation have demonstrated the profit of user feedback while ignoring the exploiting the fuzzy feedback from ordinary users. With the help of community wise and quality control, ordinary user feedback seems to be a good substitution of expert feedback. Whether or not it would service to evaluate machine translation output more user-centered and closer to expert evaluation than the state of art MT evaluation metric? How to get rid of noise, and then make use of user feedback to improve machine translation system?

Sept 27
Yutao Wang
Learning and forgetting in ASSISTments
    The talk is about a variety of machine learning approaches she has used to try to understand student learning and forgetting in ASSISTments, a web-based tutor for mathematics.

Oct 4

Oct 11
Dave Brown
"Creativity, Surprise And Design"
    How can designer surprise an evaluator with their designs? How can they surprise themselves with their own designs? Why do we care about these questions? What has surprise got to do with creativity? How does this relate to AI?

Oct 18
Hien Duong
"AI in Accounting"
** Talk postponed **
(Fall Recess - Oct 12-22)

Oct 25
Xiaolu Xiong (work he has done with Shoujing Li)
"All about ARRS: The development, problems and analysis"
    The new version of ARRS (Automatic Reassessment and Relearning System) has been officially deployed on September 1st, 2012, and it is being used by more than 150 math classes. The ARRS is an extension of the ARRS system, the propose of the ARRS system is to make sure that students well mastery the skill they have been taught, and after long period of time, they still remember and understand these skills. It works by automatically assigning students Reassessment test base on a certain schedule. And if students fail on such tests, the system will also automatically create and assign new Relearning assignments to them so that they have the opportunity of reviewing the tutoring again. Although that we have a set of defined settings to define the behavior of ARRS, and it is acceptable by most teacher who are using the system, we are still very interested in finding the set of settings which help the teaching and learning the best. We propose a serious of experiment questions to address to several important area of student learning and forgetting in the system, such as when is the best time to reassess students, which skills should we reassess them, does relearning matters, should we make ARRS smarter so it use different set of setting on different group of students. As the beginning stage of the research, we have already looked the relationship between student mastery performance and Reassessment test performance; we are expecting to have a better understanding on student short term retention and find a better setting for the first Reassessment test.

Nov 1

Nov 8
Junjie Gu & Bin Lang
topics tba

Nov 15

Nov 22
Thanksgiving Break - No Meeting

Nov 29

Dec 6
Yue Gong
"Improving an Intelligent Tutoring System Using Data Mining Techniques"
    PhD dissertation proposal

Dec 13
(hold for poss. thesis presentation)

-- Fri Dec 14 -- Last day of Semester

[Return to the WPI Homepage] [Return to the CS Homepage]

AIRG Coordinator / Wed, 26 Sep 2012 22:50:37