Cascaded Span Extraction and Response Generation for Document-Grounded Dialog

This paper summarizes our entries to both subtasks of the first DialDoc\nshared task which focuses on the agent response prediction task in\ngoal-oriented document-grounded dialogs. The task is split into two subtasks:\npredicting a span in a document that grounds an agent turn and generating an\nagent response based on a dialog and grounding document. In the first subtask,\nwe restrict the set of valid spans to the ones defined in the dataset, use a\nbiaffine classifier to model spans, and finally use an ensemble of different\nmodels. For the second subtask, we use a cascaded model which grounds the\nresponse prediction on the predicted span instead of the full document. With\nthese approaches, we obtain significant improvements in both subtasks compared\nto the baseline.\n