Corpus-based Validation of a Dialogue Model for Social Support

Title

Corpus-based Validation of a Dialogue Model for Social Support

 Author

Van der Zwaan, J.M.
Dignum, V.
Jonker, C.M.

Faculty

Technology, Policy and Management

 Department

Infrastructures, Systems and Services

 Date

2012-10-25

 Abstract

Recent developments in affective computing show that Embodied Conversational Agents (ECAs) are increasingly capable of complex social and emotional dialogues. Our research concerns the design and evaluation of an ECA that provides social support to children that are being bullied though the Internet. Recently, we proposed a domain-independent dialogue model for verbal social support. In this paper, the model is compared to actual comforting chat conversations about bullying. Analysis of the most important and complex conversation phases shows that conversation topics predicted by the model are mostly discussed in the expected phases and that conversation patterns used to discuss topics and give (verbal) social support also correspond to the model.

 To reference this document use:

http://resolver.tudelft.nl/uuid:b273ee3c-16b2-416e-ba93-944af820b180

 Source

BNAIC 2012: 24th Benelux Conference on Artificial Intelligence, Maastricht, The Netherlands, 25-26 October 2012

 Part of collection

Institutional Repository

 Document type

conference paper

 Rights

(c) 2012 Van der Zwaan, J.M.
Dignum, V.
Jonker, C.M.
Files
PDF vanderZwaan_2012.pdf 341.7 KB
Close viewer