Reva Freedman

An intelligent tutoring system that generates a natural language dialogue using dynamic multi-level planning

OBJECTIVE The objective of this research was to build an intelligent tutoring system capable of carrying on a natural language dialogue with a student who is solving a problem in physiology. Previous experiments have shown that students need practice in qualitative causal reasoning to internalize new knowledge and to apply it effectively and that they learn by putting their ideas into words. METHODS Analysis of a corpus of 75 hour-long tutoring sessions carried on in keyboard-to-keyboard style by two professors of physiology at Rush Medical College tutoring first-year medical students provided the rules used in tutoring strategies and tactics, parsing, and text generation. The system presents the student with a perturbation to the blood pressure, asks for qualitative predictions of the changes produced in seven important cardiovascular variables, and then launches a dialogue to correct any errors and to probe for possible misconceptions. The natural language understanding component uses a cascade of finite-state machines. The generation is based on lexical functional grammar. RESULTS Results of experiments with pretests and posttests have shown that using the system for an hour produces significant learning gains and also that even this brief use improves the students ability to solve problems more then reading textual material on the topic. Student surveys tell us that students like the system and feel that they learn from it. The system is now in regular use in the first-year physiology course at Rush Medical College. CONCLUSION We conclude that the CIRCSIM-Tutor system demonstrates that intelligent tutoring systems can implement effective natural language dialogue with current language technology.

Generating and Revising Hierarchical Multi-turn Text Plans in an ITS

CIRCSIM-Tutor v. 3 is a natural-language based ITS for cardiac physiology. In this paper, we describe TIPS, a new text planning engine for CIRCSIM-Tutor based on current research in text generation. Since conversations cannot be completely planned in advance, TIPS plans and executes iteratively. It maintains a goal hierarchy for the tutor while carrying on a conversation with the student. It can handle multi-turn plans on the part of the tutor, and it can back up and replan when the student gives an unexpected answer. In this paper we sketch the design of TIPS using an analysis of human-to-human tutoring transcripts to shape the requirements.

Annotation of Tutorial Dialogue Goals for Natural Language Generation.

We annotated transcripts of human tutoring dialogue for the purpose of constructing a dialogue-based intelligent tutoring system, CIRCSIM-Tutor. The tutors were professors of physiology who were also expert tutors. The students were 1st year medical students who communicated with the tutors using typed communication from separate rooms. The tutors made use of a rich variety of strategies, some specific to particular content areas and others more general, such as showing that the student holds contradictory beliefs about the domain. In this article, we describe our model of hierarchical goal structure for tutorial dialogues. We catalog each major pedagogical method we found in the dialogues, showing its structure and illustrating the features needed to represent each subgoal in its correct narrative and interpersonal context. We compare our goal structure with other analyses of tutorial dialogues.

Concrete Assignments for Teaching NLP in an M.S. Program

The professionally oriented computer science M.S. students at Northern Illinois University are intelligent, interested in new ideas, and have good programming skills and a good math background. However, they have no linguistics background, find traditional academic prose difficult and uninteresting, and have had no exposure to research. Given this population, the assignments I have found most successful in teaching Introduction to NLP involve concrete projects where students could see for themselves the phenomena discussed in class. This paper describes three of my most successful assignments: duplicating Kernighan et al.s Bayesian approach to spelling correction, a study of Greenbergs universals in the students native language, and a dialogue generation project. For each assignment I discuss what the students learned and why the assignment was successful.

Workshop on Dialog-Based Intelligent Tutoring Systems: State of the Art and New Research Directions

Within the past decade, advances in computer technology and language-processing techniques have allowed us to develop intelligent tutoring systems that feature more natural communication with students. As these dialog-based tutoring systems are maturing, there is increasing agreement on the fundamental methods that make them effective in producing learning gains. This workshop will have two goals. First, we will discuss current research the techniques that make these systems effective. Second, especially for the benefit of researchers just starting tutorial dialog projects, we will include a how-to track where experienced system-builders describe the tools and techniques that form the cores of successful systems.

ITS Tools for Natural Language Dialogue: A Domain-Independent Parser and Planner

The goal of the Atlas project is to increase the opportunities for students to construct their own knowledge by conversing (in typed form) with a natural language-based ITS. In this paper we describe two components of Atlas|APE, the integrated planning and execution system at the heart of Atlas, and CARMEL, the natural language understanding component. These components have been designed as domain-independent rule-based software, with the goal of making them both extensible and reusable. We illustrate the use of CARMEL and APE by describing Atlas-Andes, a prototype ITS built with Atlas using the Andes physics tutor as the host.

Teaching NLP to Computer Science Majors via Applications and Experiments

Most computer science majors at Northern Illinois University, whether at the B.S. or M.S. level, are professionally oriented. However, some of the best students are willing to try something completely different. NLP is a challenge for them because most have no background in linguistics or artificial intelligence, have little experience in reading traditional academic prose, and are unused to open-ended assignments with gray areas. In this paper I describe a syllabus for Introduction to NLP that concentrates on applications and motivates concepts through student experiments. Core materials include an introductory linguistics textbook, the Jurafsky and Martin textbook, the NLTK book, and a Python textbook.

Comparison of Expert Tutors Through Syntactic Analysis of Transcripts

In this paper we show that the C4.5 machine learning algorithm, applied to a number of syntactic features in transcripts, can be used to accurately differentiate between two expert human tutors. Although these tutors had taught together for years and explicitly discussed their tutoring style with one other, an analysis based on frequency of parts of speech and higher-level syntactic constructs was able to easily separate their productions.

Reducing Stereotypes of Women in Technology Through Analysis of Videogame Blog Entries

In this paper we analyze some of the most frequent stereotypes about women found in videogames. One way to negate these beliefs is to look at real data. Since contemporary students prefer blog entries to hardcopy or longer articles, in this paper we examine some inaccurate but widely held beliefs found in the videogame industry and analyze blog entries that can be used to negate them. This idea grew out of a mixed-gender class in game programming at Northern Illinois University in Spring 2012.

Improving Student Learning While Converting a Computer Architecture Course to Online Format

A required Computer Architecture course for Computer Science majors was converted to online form in Spring 2011. In this paper we discuss the changes made to make the course successful online, especially in content preparation, course organization, and the construction and handling of assignments. We discuss why we feel the revised course produces improved student learning in terms of the basic principles of scaffolding, self-explanation and multimodal learning. We also discuss how we made the course practical to administer on an ongoing basis. We hope that this experience will be helpful to other faculty members planning to convert courses in Computer Science or Engineering to an online format.