Paper: Improving interpretation robustness in a tutorial dialogue system

ACL ID W13-1738
Title Improving interpretation robustness in a tutorial dialogue system
Venue Innovative Use of NLP for Building Educational Applications
Session
Year 2013
Authors

We present an experiment aimed at improv- ing interpretation robustness of a tutorial dia- logue system that relies on detailed semantic interpretation and dynamic natural language feedback generation. We show that we can improve overall interpretation quality by com- bining the output of a semantic interpreter with that of a statistical classifier trained on the subset of student utterances where seman- tic interpretation fails. This improves on a pre- vious result which used a similar approach but trained the classifier on a substantially larger data set containing all student utterances. Fi- nally, we discuss how the labels from the sta- tistical classifier can be integrated effectively with the dialogue system?s existing error re- covery policies.