Combining Competing Language Understanding Approaches in an Intelligent Tutoring System

When implementing a tutoring system that attempts a deep understanding of students’ natural language explanations, there are three basic approaches to choose between; symbolic, in which sentence strings are parsed using a lexicon and grammar; statistical, in which a corpus is used to train a text classifier; and hybrid, in which rich, symbolically produced features supplement statistical training. Because each type of approach requires different amounts of domain knowledge preparation and provides different quality output for the same input, we describe a method for heuristically combining multiple natural language understanding approaches in an attempt to use each to its best advantage. We explore two basic models for combining approaches in the context of a tutoring system; one where heuristics select the first satisficing representation and another in which heuristics select the highest ranked representation.

[1]  InterpretationCarolyn P. Ros A Framework for Robust Semantic Interpretation , 2000 .

[2]  Carolyn Penstein Rosé,et al.  A Hybrid Text Classification Approach for Analysis of Student Essays , 2003, HLT-NAACL 2003.

[3]  Kurt VanLehn,et al.  Abductive Theorem Proving for Analyzing Student Explanations to Guide Feedback in Intelligent Tutoring Systems , 2004, Journal of Automated Reasoning.

[4]  Adam Cheyer,et al.  The Open Agent Architecture , 1997, Autonomous Agents and Multi-Agent Systems.

[5]  Patrick Pantel,et al.  Discovery of inference rules for question-answering , 2001, Natural Language Engineering.

[6]  Neil A. Macmillan,et al.  Detection Theory: A User's Guide , 1991 .

[7]  Carolyn Penstein Rosé,et al.  The Architecture of Why2-Atlas: A Coach for Qualitative Physics Essay Writing , 2002, Intelligent Tutoring Systems.

[8]  Guillermo Ricardo Simari,et al.  Multiagent systems: a modern approach to distributed artificial intelligence , 2000 .

[9]  Andrew McCallum,et al.  A comparison of event models for naive bayes text classification , 1998, AAAI 1998.

[10]  Steven Pinker,et al.  Lexical and Conceptual Semantics , 1991 .

[11]  Arthur C. Graesser,et al.  Is it an Agent, or Just a Program?: A Taxonomy for Autonomous Agents , 1996, ATAL.

[12]  Claus Zinn,et al.  A 3-Tier Planning Architecture for Managing Tutorial Dialogue , 2002, Intelligent Tutoring Systems.

[13]  Steven P. Abney Partial parsing via finite-state cascades , 1996, Natural Language Engineering.

[14]  Kurt VanLehn,et al.  The acquisition of qualitative physics knowledge during textbook-based physics training , 1997 .

[15]  Kurt VanLehn,et al.  Abductive Theorem Proving for Analyzing Student Explanations to Guide Feedback in Intelligent Tutoring Systems , 2004 .

[16]  Michael Strube,et al.  The Influence of Minimum Edit Distance on Reference Resolution , 2002, EMNLP.

[17]  Tuomas Sandholm,et al.  Distributed rational decision making , 1999 .

[18]  Christoph Walther,et al.  A Many-Sorted Calculus Based on Resolution and Paramodulation , 1982, IJCAI.

[19]  Ray Jackendoff Semantics and Cognition , 1983 .

[20]  Vincent Aleven,et al.  A Tutorial Dialogue System with Knowledge-Based Understanding and Classification of Student Explanations , 2001 .

[21]  Vincent Aleven,et al.  Pilot-Testing a Tutorial Dialogue System That Supports Self-Explanation , 2002, Intelligent Tutoring Systems.

[22]  Dekang Lin,et al.  Dependency-Based Evaluation of Minipar , 2003 .

[23]  Peter A. Flach The Geometry of ROC Space: Understanding Machine Learning Metrics through ROC Isometrics , 2003, ICML.

[24]  Markku Turunen,et al.  Adaptive Dialogue Systems - Interaction with Interact , 2002, SIGDIAL Workshop.

[25]  Claire Gardent,et al.  Improving Machine Learning Approaches to Coreference Resolution , 2002, ACL.

[26]  K. VanLehn,et al.  Overcoming the Knowledge Engineering Bottleneck for Understanding Student Language Input , 2003 .