Generating Tailored, Comparative Descriptions in Spoken Dialogue
Johanna D. Moore, Mary Ellen Foster, Oliver Lemon, Michael White · 2004
We describe an approach to presenting information in spoken dialogues that for the first time brings together multi-attribute decision models, strategic content plan-ning, state-of-the-art dialogue management, and real-ization which incorporates prosodic features. The sys-tem selects the most important subset of available op-tions to mention and the attributes that are most relevant to choosing between them, based on the user model. It also determines how to organize and express descrip-tions of selected options and attributes, including deter-mination of information structure and rhetorical struc-ture at the level of content planning, resulting in de-scriptions which, we hypothesize, are both memorable and easy for users to understand.