Tutorial Dialogue

Research has shown that human tutors teach math to students much better than does any current form of technology, whether it is simply flash cards or highly elaborate systems to teach conceptual reasoning beyond problem solving. One reason for this may be the use of language itself as a teaching medium. The local tutorial dialogue component aims to explore this possibility by pairing a language-based computer tutor with a computer-based solver and reasoner for differentiating equations. Thus the student should be able not only to attempt to solve equations, but to do so in a language-supported environment where she can ask for help, indicate she wants to work on a particular subexpression within a  larger formula, and receive natural, text-based feedback that helps her solve her individual problem rather than just generic advice.       

click images to enlarge

Current tutorial systems for math do not support language-based interactions on both the student's and tutor's part.  Additionally, many other tutorial systems do not allow the student to solve arbitrary differential equations.  Finally, although the resources needed to implement both traditional tutoring systems and our language-based system are roughly the same, future versions of our system should be much more scalable.