Integrating breakdown detection into dialogue systems to improve knowledge management: encoding temporal utterances with memory attention

Seolhwa Lee, Dongyub Lee, Danial Hooshyar, Jaechoon Jo, Heuiseok Lim

Research output: Contribution to journalArticle

Abstract

With the increasing pervasiveness of smart phones and smart devices, dialogue systems are gaining ever growing attention from both academic and industry. These systems can be broadly classified into two categories, one that is aimed at helping user to gain new knowledge and one that can chat with users without completing any specific tasks. Although dialogue systems are improving substantially, the user experience of such systems are still unsatisfactory as there are no specific rules covering all possible situations of real human–machine dialogue, resulting in breakdowns. There are two technical issues affecting the detection of dialogue breakdown in an open domain conversation: human resources to prepare and annotate a large chunk of conversation data and dialogue histories containing words that don’t appear directly in training data. To tackle these issues, we propose a novel encoding method for temporal utterances with memory attention based on end-to-end dialogue breakdown detection. Specifically, long short-term memory (LSTM) is employed to encode each word of all previous user and system utterances. Encoded vectors from LSTM (user and system utterances), along with system and user utterances from sentence embedding, are then stored in memory wherein an attention mechanism is applied to select the most relevant piece of words from system and user utterances for breakdown detection. An evaluation of the proposed approach on a breakdown detection task (DBDC3) showed that the model for single-labeled breakdown detection outperforms other state-of-the-art methods in a classification task. In conclusion, a more effective knowledge gain and management can be achieved by integration of our proposed breakdown detection into dialogue systems.

Original languageEnglish
JournalInformation Technology and Management
DOIs
Publication statusAccepted/In press - 2019 Jan 1

Fingerprint

Knowledge management
knowledge management
communication technology
Data storage equipment
dialogue
Personnel
conversation
Industry
chat
Long short-term memory
Breakdown
human resources
industry
history
evaluation
management
knowledge
experience

Keywords

  • Breakdown detection
  • Dialogue system
  • Knowledge management
  • Memory attention

ASJC Scopus subject areas

  • Information Systems
  • Communication
  • Business, Management and Accounting (miscellaneous)

Cite this

Integrating breakdown detection into dialogue systems to improve knowledge management : encoding temporal utterances with memory attention. / Lee, Seolhwa; Lee, Dongyub; Hooshyar, Danial; Jo, Jaechoon; Lim, Heuiseok.

In: Information Technology and Management, 01.01.2019.

Research output: Contribution to journalArticle

@article{8b2311382ba74efea98a513171635595,
title = "Integrating breakdown detection into dialogue systems to improve knowledge management: encoding temporal utterances with memory attention",
abstract = "With the increasing pervasiveness of smart phones and smart devices, dialogue systems are gaining ever growing attention from both academic and industry. These systems can be broadly classified into two categories, one that is aimed at helping user to gain new knowledge and one that can chat with users without completing any specific tasks. Although dialogue systems are improving substantially, the user experience of such systems are still unsatisfactory as there are no specific rules covering all possible situations of real human–machine dialogue, resulting in breakdowns. There are two technical issues affecting the detection of dialogue breakdown in an open domain conversation: human resources to prepare and annotate a large chunk of conversation data and dialogue histories containing words that don’t appear directly in training data. To tackle these issues, we propose a novel encoding method for temporal utterances with memory attention based on end-to-end dialogue breakdown detection. Specifically, long short-term memory (LSTM) is employed to encode each word of all previous user and system utterances. Encoded vectors from LSTM (user and system utterances), along with system and user utterances from sentence embedding, are then stored in memory wherein an attention mechanism is applied to select the most relevant piece of words from system and user utterances for breakdown detection. An evaluation of the proposed approach on a breakdown detection task (DBDC3) showed that the model for single-labeled breakdown detection outperforms other state-of-the-art methods in a classification task. In conclusion, a more effective knowledge gain and management can be achieved by integration of our proposed breakdown detection into dialogue systems.",
keywords = "Breakdown detection, Dialogue system, Knowledge management, Memory attention",
author = "Seolhwa Lee and Dongyub Lee and Danial Hooshyar and Jaechoon Jo and Heuiseok Lim",
year = "2019",
month = "1",
day = "1",
doi = "10.1007/s10799-019-00308-x",
language = "English",
journal = "Information Technology and Management",
issn = "1385-951X",
publisher = "Kluwer Academic Publishers",

}

TY - JOUR

T1 - Integrating breakdown detection into dialogue systems to improve knowledge management

T2 - encoding temporal utterances with memory attention

AU - Lee, Seolhwa

AU - Lee, Dongyub

AU - Hooshyar, Danial

AU - Jo, Jaechoon

AU - Lim, Heuiseok

PY - 2019/1/1

Y1 - 2019/1/1

N2 - With the increasing pervasiveness of smart phones and smart devices, dialogue systems are gaining ever growing attention from both academic and industry. These systems can be broadly classified into two categories, one that is aimed at helping user to gain new knowledge and one that can chat with users without completing any specific tasks. Although dialogue systems are improving substantially, the user experience of such systems are still unsatisfactory as there are no specific rules covering all possible situations of real human–machine dialogue, resulting in breakdowns. There are two technical issues affecting the detection of dialogue breakdown in an open domain conversation: human resources to prepare and annotate a large chunk of conversation data and dialogue histories containing words that don’t appear directly in training data. To tackle these issues, we propose a novel encoding method for temporal utterances with memory attention based on end-to-end dialogue breakdown detection. Specifically, long short-term memory (LSTM) is employed to encode each word of all previous user and system utterances. Encoded vectors from LSTM (user and system utterances), along with system and user utterances from sentence embedding, are then stored in memory wherein an attention mechanism is applied to select the most relevant piece of words from system and user utterances for breakdown detection. An evaluation of the proposed approach on a breakdown detection task (DBDC3) showed that the model for single-labeled breakdown detection outperforms other state-of-the-art methods in a classification task. In conclusion, a more effective knowledge gain and management can be achieved by integration of our proposed breakdown detection into dialogue systems.

AB - With the increasing pervasiveness of smart phones and smart devices, dialogue systems are gaining ever growing attention from both academic and industry. These systems can be broadly classified into two categories, one that is aimed at helping user to gain new knowledge and one that can chat with users without completing any specific tasks. Although dialogue systems are improving substantially, the user experience of such systems are still unsatisfactory as there are no specific rules covering all possible situations of real human–machine dialogue, resulting in breakdowns. There are two technical issues affecting the detection of dialogue breakdown in an open domain conversation: human resources to prepare and annotate a large chunk of conversation data and dialogue histories containing words that don’t appear directly in training data. To tackle these issues, we propose a novel encoding method for temporal utterances with memory attention based on end-to-end dialogue breakdown detection. Specifically, long short-term memory (LSTM) is employed to encode each word of all previous user and system utterances. Encoded vectors from LSTM (user and system utterances), along with system and user utterances from sentence embedding, are then stored in memory wherein an attention mechanism is applied to select the most relevant piece of words from system and user utterances for breakdown detection. An evaluation of the proposed approach on a breakdown detection task (DBDC3) showed that the model for single-labeled breakdown detection outperforms other state-of-the-art methods in a classification task. In conclusion, a more effective knowledge gain and management can be achieved by integration of our proposed breakdown detection into dialogue systems.

KW - Breakdown detection

KW - Dialogue system

KW - Knowledge management

KW - Memory attention

UR - http://www.scopus.com/inward/record.url?scp=85074105238&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85074105238&partnerID=8YFLogxK

U2 - 10.1007/s10799-019-00308-x

DO - 10.1007/s10799-019-00308-x

M3 - Article

AN - SCOPUS:85074105238

JO - Information Technology and Management

JF - Information Technology and Management

SN - 1385-951X

ER -