Information Presentation in Spoken Dialogue Systems
43856Authors
nan
Abstract
To tackle the problem of presenting a large number of options in spoken dia- logue systems, we identify compelling op- tions based on a model of user preferences, and present tradeoffs between alternative options explicitly. Multiple attractive op- tions are structured such that the user can gradually refine her request to find the optimal tradeoff. We show that our ap- proach presents complex tradeoffs under- standably, increases overall user satisfac- tion, and significantly improves the user’s overview of the available options. More- over, our results suggest that presenting users with a brief summary of the irrele- vant options increases users’ confidence in having heard about all relevant options. 1
Keywords
Full Content
Information Presentation in Spoken Dialogue Systems Vera Demberg Institute for Natural Language Processing (IMS) University of Stuttgart D-70174 Stuttgart V.Demberg@gmx.de Johanna D. Moore School of Informatics University of Edinburgh Edinburgh, EH8 9LW, GB J.Moore@ed.ac.uk Abstract To tackle the problem of presenting a large number of options in spoken dia- logue systems, we identify compelling op- tions based on a model of user preferences, and present tradeoffs between alternative options explicitly. Multiple attractive op- tions are structured such that the user can gradually refine her request to find the optimal tradeoff. We show that our ap- proach presents complex tradeoffs under- standably, increases overall user satisfac- tion, and significantly improves the user’s overview of the available options. More- over, our results suggest that presenting users with a brief summary of the irrele- vant options increases users’ confidence in having heard about all relevant options. 1 Introduction The goal of spoken dialogue systems (SDS) is to offer efficient and natural access to applica- tions and services, such as email and calendars, travel and entertainment booking, and product rec- ommendation. In evaluating nine SDS in the DARPA Communicator domain (flight, hotel, and rental car hire), Walker et al. (2001) found that (1) shorter task duration correlates with higher user satisfaction, and (2) the information presentation phase of dialogues is the primary contributor to dialogue duration. During this phase, the typical system enumerates the set of options that match the user’s constraints, as shown in Figure 1. The user can then refine these options by offering new constraints. When the number of options to be presented is large, this process can be painstaking, leading to reduced user satisfaction. Moreover, as S: I found 28 round trips from Edinburgh to Brussels on October 31st and returning Edinburgh on November 6th. For option 1, I have an outbound flight with KLM departing at 7 am and arriving at 8:49 am non-stop. A return flight is with KLM departing at 10:30 am and arriving at 11:28 am with 1 stopover. Ticket price is £165. Please say “next option”, or, “flight details”, or, “I’ll take it”. U: Next option S: For option 2, I have an outbound flight with KLM de- parting at 3:30 pm and arriving at 5:19 pm nonstop. A return flight is with KLM departing at 8:25 pm and arriving at 12:28 pm with 1 stopover. Ticket price is £165 please say “next option”, or, “flight details”, or, “I’ll take it.” U: Next option S: ... Figure 1: Typical Information Presentation Phase of a Communicator Dialogue Walker et al. (2004) observe, having to access the set of available options sequentially makes it diffi- cult for the user to remember the various aspects of multiple options and to compare them in memory. Clearly, alternative strategies to sequential pre- sentation of information in SDS are needed. Re- cently, two approaches have been proposed. In the user-model (UM) based approach, the sys- tem identifies a small number of options that best match the user’s preferences (Moore et al., …
Content truncated to 500 words
Labeled Text
To tackle the problem of presenting a large number of options in spoken dialogue systems, we identify compelling options based on a model of user preferences, and present tradeoffs between alternative options explicitly. Multiple attractive options are structured such that the user can gradually refine her request to find the optimal tradeoff. We show that our approach presents complex tradeoffs understandably, increases overall user satisfaction, and significantly improves the user’s overview of the available options. Moreover, our results suggest that presenting users with a brief summary of the irrelevant options increases users’ confidence in having heard about all relevant options. 1
Publication Details
- Paper ID:
- 43856
- Year:
- 2006
- Publication:
- TIPSTER
- Publication Type:
- conf
- DOI:
- nan