Computer Science Department

AIRG Topics - Spring 2009

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

Jan 22

No meeting this week

Jan 29

No meeting this week

Feb 05

Dovan Rai and Yue Gong

"Motivational effect in learning: an empirical analysis"

In the field of educational systems, there has been considerable work in detecting the affective state of students as higher motivation is believed to give better learning and enhancing motivation has been a major pursuit. Intelligent Tutoring Systems (ITS) researchers are also becoming increasingly aware of making the learning process more motivating. In this study, we are interested in using empirical analyses to estimate the impact of motivation on learning.

We used a model of motivation that divides it into four components---interest, effort, confidence and satisfaction---and compared their relative impact on learning. Based on a questionnaire survey, we grouped students into high and low categories of motivation and its components. We estimated knowledge tracing parameters for each subgroup using a Dynamic Bayesian network. We analyzed the learning rates for each group and found that motivation and all its components have a varying but positive effect on learning. We also calculated the number of practice opportunities required to master a skill for each subgroup, and found that the low motivation group required 40 practice opportunities while the high motivation group only needed 28 opportunities.

We are trying to create a paradigm to measure and compare learning in different motivational subgroups. This can give us insight about interrelations among motivation, motivational components and learning, in order to help us build more efficient or better motivating tutors.

Feb 12

Zach Pardos

"A Method for Detecting Learning Effects of Items In a Randomized Problem Set"

Researchers that make tutoring systems would like to know which pieces of educational content are most effective at promoting learning among their students. Randomized controlled experiments are often used to determine which content produces more learning in an ITS. While these experiments are powerful they are often very costly to setup and run. The majority of data collected in many ITS systems consist of answers to a finite set of questions of a given skill often presented in a random sequence. We propose a Bayesian method to detect which questions produce the most learning in this random sequence of data.

We confine our analysis to random sequences with two, three and four questions. Two simulation studies were run to investigate the validity of the method and boundaries on what learning probability differences could be reliably detected with various numbers of users. Finally, real tutor data from random sequence problem sets was analyzed. Results of the simulation data analysis showed that the method reported reliability (p < 0.05) in 56 of 112 simulation experiments with less than 5% error. In the analysis of real student data, the method returned statistically reliable choices of best question in four out of seven problem sets.

Feb 19

Academic Advising Appointment Day: No meeting

Feb 26

Elijah Forbes-Summers

"Determining the effect of gaming behavior on learning using a Bayesian network model"

Yue Gong and I recently began a follow-up study to her earlier work with Dovan Rai. Previously Dovan and Yue studied the effect of motivation on learning utilizing a Bayesian net model that was based on the model presented in Joseph Beck's work "Does help help? Introducing the Bayesian Evaluation and Assessment methodology". In our study we will be using a similar model to investigate whether students gaming of the tutoring system affects their learning. As this project is still being planned and implemented this will be a short talk (~20 minutes) in order to receive feedback and suggestions from other AIRG members. I will outline the study, discuss briefly the aspects of Assistments that are important to our study, and describe the Bayesian model and the methods for estimating gaming behavior that we use.

Mar 05

Mike Sao Pedro

Postponed

Mar 12

Spring Recess

Mar 19

No meeting

Mar 26

Mingyu Feng

"Using learning decomposition to analyze instructional effectiveness in the ASSISTment system"

A basic question of instruction is how effective it is in promoting student learning. This paper presents a study determining the relative efficacy of different instructional content by applying an educational data mining technique, learning decomposition. We use logistic regression to determine how much learning caused by different methods of presenting same skill, relative to each other. We analyze more than 60,000 performance data across 181 items from more than 2,000 students. Our results show that items are not all as effective on promoting student learning. We also did preliminary study on validating our results by comparing them with rankings from human experts. Our study demonstrates an easier and quicker approach of evaluating the quality of ITS contents than experimental studies.

Apr 02

Leena Razzaq

"Adapting tutoring strategies to students"

Tutoring systems often rely on interactive tutored problem solving to help students learn math, which requires students to work through problems step-by-step while the system provides help and feedback. This approach has been shown to be effective in improving student performance in numerous studies. However, tutored problem solving may not be the most effective approach for all students. In previous studies, we found that tutored problem solving was more effective than less interactive approaches, such as simply presenting a worked out solution, for students who were not proficient in math. More proficient students benefited more from seeing solutions rather than going through all of the steps. Can we use the results of these studies to adapt the tutoring strategy to the students' math proficiency? In this presentation, I will talk about an experiment to examine whether adapting the tutoring to the student's proficiency is better than randomly assigning students to a condition.

Apr 9

Keith Pray

Sleep Data Exploration - Results Update

Sleep disorder studies offer a rich data set which includes: time sequence data in the form of electroencephalograms, electromyograms, electrooculograms, electrocardiograms, blood oxygen levels, body position, and blood pressure; data collected in questionnaires on day time sleepiness and depression which can vary in accuracy and interpretation from patient to patient; more reliable data on patient demographics such as smoking and drinking habits, and medical history; data summarized by technicians from the time sequence data and classifications of sleep disorder severity. In the interest of providing ways in which to use the time sequence data directly in the construction of models for predicting sleep disorder severity a set of exploratory experiments are being conducted to characterize the predictive power of the non-time sequence data. In particular the questionnaire and demographic data are examined in relationship to the Beck Depression Index.

Apr 16

Amro Khasawneh

Human Sleep Data Analysis Using Hidden Markov Models

This talk describes an ongoing exploratory analysis of a large human sleep database. We focus on prediction of patient sleep disorder severity based on polysomnographic time sequence signals, in particular EEG. Hidden Markov Models are used to describe the underlying process that generates the EEG time series, aiming to capture temporal patterns specific to each severity level on a mild, moderate and severe scale. Ongoing experimental results are presented.

Apr 23

Project Presentation Day

Apr 30

Yu Guo

title & abstract tbd

May 1

Last day of Semester

AIRG Coordinator / 6 April 09