Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

Entrainment in Human-to-Human Dialogue and its Application in End-to-End Dialogue Systems

Yu, Mingzhi (2022) Entrainment in Human-to-Human Dialogue and its Application in End-to-End Dialogue Systems. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

[img]
Preview
PDF (FINAL VERSION)
Published Version

Download (3MB) | Preview

Abstract

Entrainment is a linguistic phenomenon in which people mimic each other in their conversations. It occurs in a wide range of linguistic dimensions. Entrainment has been exploited in various natural language processing tasks related to dialogue, such as dialogue outcome prediction and dialogue response generation. However, only a few studies have attempted to incorporate entrainment into neural network-based dialogue systems systematically. The present thesis aims to build a neural network-based end-to-end response generation model capable of generating diverse responses by leveraging lexical entrainment, a type of entrainment based on text features. We first demonstrate an automatic entrainment measure relying on conventional similarity metrics based on a bag-of-words approach. Then we show an alternative neural network-based approach to perform the same core similarity measure for entrainment quantification. Lastly, we proposed an end-to-end dialogue response generation model that controls entrainment degree to aid response diversity. We will focus on investigating the effect of incorporating lexical entrainment in the end-to-end dialogue response generation model.


Share

Citation/Export:
Social Networking:
Share |

Details

Item Type: University of Pittsburgh ETD
Status: Unpublished
Creators/Authors:
CreatorsEmailPitt UsernameORCID
Yu, Mingzhimiy39@pitt.edumiy39
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairLitman, Dianedlitman@pitt.edudlitman
Committee MemberWalker, ErinEAWALKER@pitt.eduEAWALKER
Committee MemberKovashka, Adrianakovashka@cs.pitt.edukovashka
Committee MemberMa, Shuangshuama@microsoft.com
Date: 17 January 2022
Date Type: Publication
Defense Date: 5 November 2021
Approval Date: 17 January 2022
Submission Date: 30 November 2021
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Number of Pages: 132
Institution: University of Pittsburgh
Schools and Programs: School of Computing and Information > Computer Science
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: Entrainment, Dialogue system, Dialogue response generation, Neural network model
Date Deposited: 17 Jan 2022 15:01
Last Modified: 17 Jan 2022 15:01
URI: http://d-scholarship.pitt.edu/id/eprint/41983

Metrics

Monthly Views for the past 3 years

Plum Analytics


Actions (login required)

View Item View Item