Someone (must have) said (sometime or other): "An explanation is a logical proof of the correctness of an answer."
Presumably, you disagree with this both as a statement about explanations provided by humans and about the kind of explanations that AI programs should provide. Discuss your position thoroughly. If an AI program is to provide better explanations, what problems need to be addressed? How would you address them?
The concept of coherence(consistency) arises repeatedly in your reading list. What is coherence? Does it mean the same thing in the various topics? What problems are associated with modeling coherence?
Present an otuline of the components you believe are needed in the following expert models to make them powerful in driving intelligent argumentation skills learning environments: (a) for public policy argumentation; (b) for scientific argumentation. Pick either and discuss the possible uses for student modeling.
The literature on NL processing contains different accounts of discourse structure. What does each theory attempt to account for? Are the accounts incompatible or not and how? Which do you think is most useful in building a system that can participate in cooperative argument?
R. Michael Young < myoung+@pitt.edu>