Davide Fossati

Supporting Computer Science Curriculum: Exploring and Learning Linked Lists with iList

We developed two versions of a system, called iList, that helps students learn linked lists, an important topic in computer science curricula. The two versions of iList differ on the level of feedback they can provide to the students, specifically in the explanation of syntax and execution errors. The system has been fielded in multiple classrooms in two institutions. Our results indicate that iList is effective, is considered interesting and useful by the students, and its performance is getting closer to the performance of human tutors. Moreover, the system is being developed in the context of a study of human tutoring, which is guiding the evolution of iList with empirical evidence of effective tutoring.

Aggregation Improves Learning: Experiments in Natural Language Generation for Intelligent Tutoring Systems

To improve the interaction between students and an intelligent tutoring system, we developed two Natural Language generators, that we systematically evaluated in a three way comparison that included the original system as well. We found that the generator which intuitively produces the best language does engender the most learning. Specifically, it appears that functional aggregation is responsible for the improvement.

A Mixed Trigrams Approach for Context Sensitive Spell Checking

This paper addresses the problem of real-word spell checking, i.e., the detection and correction of typos that result in real words of the target language. This paper proposes a methodology based on a mixed trigrams language model. The model has been implemented, trained, and tested with data from the Penn Treebank. The approach has been evaluated in terms of hit rate, false positive rate, and coverage. The experiments show promising results with respect to the hit rates of both detection and correction, even though the false positive rate is still high.

Learning Linked Lists: Experiments with the iList System

This paper presents the first experiments with an Intelligent Tutoring System in the domain of linked lists, a fundamental topic in Computer Science. The system has been deployed in an introductory college-level Computer Science class, and engendered significant learning gains. A constraint-based approach has been adopted in the design and implementation of the system. We describe the system architecture, its current functionalities, and the future directions of its development.

Towards improving programming habits to create better computer science course outcomes

We examine a large dataset collected by the Marmoset system in a CS2 course. The dataset gives us a richly detailed portrait of student behavior because it combines automatically collected program snapshots with unit tests that can evaluate the correctness of all snapshots. We find that students who start earlier tend to earn better scores, which is consistent with the findings of other researchers. We also detail the overall work habits exhibited by students. Finally, we evaluate how students use release tokens, a novel mechanism that provides feedback to students without giving away the code for the test cases used for grading, and gives students an incentive to start coding earlier. We find that students seem to use their tokens quite effectively to acquire feedback and improve their project score, though we do not find much evidence suggesting that students start coding particularly early.

The Role of Positive Feedback in Intelligent Tutoring Systems

The focus of this study is positive feedback in one-on-one tutoring, its computational modeling, and its application to the design of more effective Intelligent Tutoring Systems. A data collection of tutoring sessions in the domain of basic Computer Science data structures has been carried out. A methodology based on multiple regression is proposed, and some preliminary results are presented. A prototype Intelligent Tutoring System on linked lists has been developed and deployed in a collegelevel Computer Science class.

The problem of ontology alignment on the web: a first report

This paper presents a general architecture and four algorithms that use Natural Language Processing for automatic ontology matching. The proposed approach is purely instance based, i.e., only the instance documents associated with the nodes of ontologies are taken into account. The four algorithms have been evaluated using real world test data, taken from the Google and LookSmart online directories. The results show that NLP techniques applied to instance documents help the system achieve higher performance.

Generating proactive feedback to help students stay on track

In a tutoring system based on an exploratory environment, it is also important to provide direct guidance to students We endowed iList, our linked list tutor, with the ability to generate proactive feedback using a procedural knowledge model automatically constructed from the interaction of previous students with the system We compared the new version of iList with its predecessors and human tutors Our evaluation shows that iList is effective in helping students learn.

Affect detection from non-stationary physiological data using ensemble classifiers

Affect detection from physiological signals has received considerable attention. One challenge is that physiological measures exhibit considerable variations over time, making classification of future data difficult. The present study addresses this issue by providing insights on how diagnostic physiological features of affect change over time. Affective physiological data (electrocardiogram, electromyogram, skin conductivity, and respiration) was collected from four participants over five sessions each. Classification performance of a number of training strategies, under different conditions of features selection and engineering, were compared using an adaptive classifier ensemble algorithm. Analysis of the performance of individual physiological channels for affect detection is also provided. The key result is that using pooled features set for affect detection is more accurate than using day-specific features. A decision fusion strategy which combines decisions from classifiers trained on individual channels data outperformed a features fusion strategy. Results also show that the performance of the ensemble is affected by the choice of the base classifier and the alpha factor used to update the member classifiers of the ensemble. Finally, the corrugator and zygomatic facial EMGs were found to be more reliable measures for detecting the valence component of affect compared to other channels.

A Scalable Intelligent Tutoring System Framework for Computer Science Education

Computer Science is a difficult subject with many fundamentals to be taught, usually involving a steep learning curve for many students. It is some of these initial challenges that can turn students away from computer science. We have been developing a new Intelligent Tutoring System, ChiQat-Tutor, that focuses on tutoring of Computer Science fundamentals. Here, we outline the system under development, while bringing particular attention to its architecture and how it attains the primary goals of being easily extensible and providing a low barrier of entry to the end user. The system is broadly broken down into lessons, teaching strategies, and utilities, which work together to promote seamless integration of components. We also cover currently developed components in the form of a case study, as well as detailing our experience of deploying it to an undergraduate Computer Science classroom, leading to learning gains on par with prior work.