The final goal of the DIALOG project is a natural tutorial dialog
between a student and an assistance system for mathematics. Based on the
research of the previous project phase, we will concentrate on novel and
deep problems regarding the interpretation and evaluation of proof step
utterances. The most important problems we shall concentrate on in the
final funding period are:
the linguistic analysis combining deep and shallow processing methods
for processing informal input to the tutoring system,
the evaluation of user utterances describing proof steps with respect
to soundness, granularity, and relevance, and
the disambiguation of ambiguous utterances using a combination of
discourse and domain knowledge and underspecification techniques.
We expect the results of our research to be highly relevant for both
dialog processing and mathematical knowledge management. Furthermore,
the results obtained in this project, using mathematics as a case study,
may be generalized easily to dialogs on other domains of technical
nature.
