Different ways of ending human-machine dialogues

• More and more dialogue systems are developed to provide public services to inexperienced users. This makes the competition harder among developers and increases the interest in finding methods to evaluate and improve the overall performance of dialogue systems. The aim of this study is to propose a method to evaluate users’ satisfaction in dialogue interactions. The hypothesis being tested is that it is possible to derive information about the users’ satisfaction with the interaction, by simply analysing the way in which users ends the dialogues. The users' final utterances were analysed in order to classify the type of utterance and to find out some prosodic correlates that could signal the users' attitude towards the system. To test this hypothesis we carried out a preliminary analysis of the way in which users' of two Swedish multi-modal dialogue systems (namely August and Adapt) ended their interactions.

Ämnesord

HUMANIORA  -- Språk och litteratur -- Studier av enskilda språk (hsv//swe)

HUMANITIES  -- Languages and Literature -- Specific Languages (hsv//eng)

NATURVETENSKAP  -- Data- och informationsvetenskap -- Datavetenskap (hsv//swe)

NATURAL SCIENCES  -- Computer and Information Sciences -- Computer Sciences (hsv//eng)

SAMHÄLLSVETENSKAP  -- Medie- och kommunikationsvetenskap -- Kommunikationsvetenskap (hsv//swe)

SOCIAL SCIENCES  -- Media and Communications -- Communication Studies (hsv//eng)

Nyckelord

conversation

human-machine

dialogue

Languages and linguistics

Språkvetenskap

Computer science

Datalogi

Human communication

Kommunikation mellan människor

Publikations- och innehållstyp

ref (ämneskategori)

kon (ämneskategori)