Towards a response selection system for spoken requests in a physical domain

Andisheh Partovi, Ingrid Zukerman, Quan Tran

Research output: Chapter in Book/Report/Conference proceedingConference PaperResearchpeer-review

Abstract

In this paper, we introduce a corpus comprising requests for objects in physical spaces, and responses given by people to these requests. We generated two datasets based on this corpus: A manually-tagged dataset, and a dataset which includes features that are automatically extracted from the output of a Spoken Language Understanding module. These datasets are used in a classification-based approach for generating responses to spoken requests. Our results show that, surprisingly, classifiers trained on the second dataset outperform those trained on the first, and produce acceptable levels of performance.

Original languageEnglish
Title of host publicationProceedings of the Linguistic and Cognitive Approaches To Dialog Agents Workshop co-located with the 26th International Joint Conference on Artificial Intelligence (IJCAI 2017), Melbourne, Australia, August 21, 2017
EditorsRafal Rzepka, Jordi Vallverdu, Andre Wlodarczyk
Place of PublicationAachen Germany
PublisherRheinisch-Westfaelische Technische Hochschule Aachen
Pages20-27
Number of pages8
Publication statusPublished - 2017
EventLinguistic and Cognitive Approaches To Dialog Agents Workshop 2017 - Melbourne, Australia
Duration: 21 Aug 201721 Aug 2017
http://ceur-ws.org/Vol-1926/

Workshop

WorkshopLinguistic and Cognitive Approaches To Dialog Agents Workshop 2017
Abbreviated titleLaCATODA 2017
CountryAustralia
CityMelbourne
Period21/08/1721/08/17
Internet address

Cite this