• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.11a
• /
• pp.308-310
• /
• 2022

With the rise of the Al wave, task-oriented dialogue systems have become one of the popular research directions in academia and industry. Currently, task-oriented dialogue systems mainly adopt pipelined form, which mainly includes natural language understanding, dialogue state decision making, dialogue state tracking and natural language generation. However, pipelining is prone to error propagation, so many task-oriented dialogue systems in the market are only for single-round dialogues. Usually single- domain dialogues have relatively accurate semantic understanding, while they tend to perform poorly on multi-domain, multi-round dialogue datasets. To solve these issues, we developed a paddle-based multi-domain task-oriented Chinese dialogue system. It is based on NEZHA-base pre-training model and CrossWOZ dataset, and uses intention recognition module, dichotomous slot recognition module and NER recognition module to do DST and generate replies based on rules. Experiments show that the dialogue system not only makes good use of the context, but also effectively addresses long-term dependencies. In our approach, the DST of dialogue tracking state is improved, and our DST can identify multiple slotted key-value pairs involved in the discourse, which eliminates the need for manual tagging and thus greatly saves manpower.

• Journal of Korean language education
• /
• v.29 no.2
• /
• pp.73-100
• /
• 2018

The purpose of this study is to investigate learner-learner dialogue during speaking tasks. In the Korean language classroom, conversation between learners is an important activity as speaking practice. However, learner dialogue is also a tool to enable learners to collaboratively conduct various cognitive activities in the classroom. In previous research, it was unfolded that through learner-learner dialogue, learners can solve second-language related problems and set a goal to carry out tasks. Therefore, this study analyzed learner-learner dialogue to investigate what kinds of cognitive activities are activated during the role-play task. As a result, the learners collaboratively generated and monitored language and content for role play. Also, in order to accomplish tasks more successfully, learners shared the same understanding about the goal of the task, and tried to manage the task procedure. Through learner-learner dialogue, learners can participate in cognitive activities such as content, language construction, and task management voluntarily without the help from teachers. This means that learner-learner dialogue can be an activity to support language learning tasks. Also, it can make learners actively involved in learning and by sharing resources with each other. It is also important that learners can experience language use that participates in real-world communication activities, such as learning in the classroom and collaborating with peer learners. This study is an exploratory study for a basic understanding of learner's conversation as a cognitive activity, and the scope of the study is limited to clarifying contents of learner-learner dialogue as a cognitive activity in speaking tasks. Based on the findings of this study, future research should be conducted on the function of learner-learner dialogue as a cognitive activity in Korean language learning and its role in the classroom of Korean language education.

• Design Convergence Study
• /
• v.18 no.3
• /
• pp.71-87
• /
• 2019

Previous research has shown that the use of self-disclosure, the process of revealing personal thoughts and feelings, in conversational agents (CAs) increases overall user evaluations. However, research exploring the effects of self-disclosure in different situations or dialogue types is limited. This study investigated the effects of self-disclosure and dialogue type (task- vs. social-oriented) on trust, usefulness, and usage intention. Results showed significant interaction effects between self-disclosure and dialogue type. For CAs that did not use self-disclosure, trust, usefulness, and usage intention were higher in task-oriented dialogues. In contrast, CAs that did use self-disclosure had higher trust, usefulness, and usage intention in social-oriented dialogues. These results suggest that researchers and designers should consider the specific dialogue types and corresponding user goals when adding human qualities, such as self-disclosure, to CAs.

• Journal of Broadcast Engineering
• /
• v.24 no.1
• /
• pp.58-66
• /
• 2019

Recognition of named entity such as proper nouns in conversation sentences is the most fundamental and important field of study for efficient conversational information prediction. The most important part of a task-oriented dialogue system is to recognize what attributes an object in a conversation has. The named entity recognition model carries out recognition of the named entity through the preprocessing, word embedding, and prediction steps for the dialogue sentence. This study aims at using user - defined dictionary in preprocessing stage and finding optimal parameters at word embedding stage for efficient dialogue information prediction. In order to test the designed object name recognition model, we selected the field of daily chemical products and constructed the named entity recognition model that can be applied in the task-oriented dialogue system in the related domain.

• Korean Journal of Cognitive Science
• /
• v.17 no.2
• /
• pp.139-150
• /
• 2006

Speech recognition errors cause fatal results in a spoken dialogue system. When a system can not determine the speech-act of u utterance due to speech recognition errors, a dialogue system has a difficulty in continuing conversation. In this paper, we propose strategies for sub-dialogue generation by inferring the speech-act of an utterance with patterns of recognition errors on the field of form-filling dialogue. We used the proposed method on a plan-based dialogue model, corrected 27% of incomplete tasks, and acquired overall 89% of task completion rate.

• KIPS Transactions on Software and Data Engineering
• /
• v.6 no.11
• /
• pp.499-506
• /
• 2017

Chatter bots are normally used in task-oriented dialogue systems to support free conversations. However, there is not much research on how chatter bots as auxiliary system should be different from independent ones. In this paper, we have developed a chatter bot for a dialogue-based computer assisted language learning (DB-CALL) system. We compared the chatter bot in two different cases: as an independent bot, and as an auxiliary system. The results showed that, the chatter bot as an auxiliary system showed much lower satisfaction than the independent one. A discussion is held about the difference between an auxiliary chatter bot and an independent bot. In addition, we evaluated a search-based chatter bot and a deep learning based chatter bot. The advantages and disadvantages of both methods are discussed.

• Proceedings of the KSPS conference
• /
• 2007.05a
• /
• pp.330-333
• /
• 2007

This paper presents a dialogue interface using the dialogue management system as a method for controlling home appliances in Home Network Services. In order to realize this type of dialogue interface, we annotated 96,000 utterance pair sized dialogue set and developed an example-based dialogue system. This paper introduces the automatic error correction module for the SMS-styled sentence. With this module we increase the accuracy of NLU(Natural Language Understanding) module. Our NLU module shows an accuracy of 86.2%, which is an improvement of 5.25% over than the baseline. The task completeness of the proposed SMS dialogue interface was 82%.

• Journal of Broadcast Engineering
• /
• v.27 no.5
• /
• pp.685-693
• /
• 2022

The Metaverse is getting popular due to the demands for digital transformation and non-contact communication platforms. A conversation system which facilitates communication is not widely applied yet in Metaverse. In this work, we present a method that revises primitive dialogue state using circumstance information from Metaverse. The presented model that leverages both dialogue and circumstance information consists of a dialogue state tracking module and a circumstance state tracking module. In the model, a dialogue state is updated with an algorithm which compares a dialogue state and a circumstance state. As a conversation that reaffirms user intent is added, a wrong dialogue state can be revised and the accuracy of a conversation system can be improved.

• Proceedings of the KSPS conference
• /
• 2006.11a
• /
• pp.109-112
• /
• 2006

This paper presents a dialogue interface using the dialogue management system as a method for controlling home appliances in Home Network Services. In order to realize this type of dialogue interface, we first investigated the user requirements for Home Network Services by analyzing the dialogues entered by users. Based on the analysis, we were able to extract 15 user intentions and 22 semantic components. In our study, example dialogues were collected from WOZ (Wizard-of-OZ) environment to implement a reasoning model for generating meaningful responses for example-based dialogue modeling technique. An overview of the Home Network Control System using proposed dialogue interface will be presented. Lastly, we will show that the Dialogue Management System trained with our collected dialogues behaves properly to achieve its task of controlling Home Network appliances by going through the steps of natural language understanding, response reasoning, response generation.

• Annual Conference on Human and Language Technology
• /
• 2017.10a
• /
• pp.56-59
• /
• 2017

CNN(Convolutional Neural Network)을 이용하여 발화 주제 다중 분류 task를 multi-labeling 방법과, cluster 방법을 이용하여 수행하고, 각 방법론에 MSE(Mean Square Error), softmax cross-entropy, sigmoid cross-entropy를 적용하여 성능을 평가하였다. Network는 음절 단위로 tokenize하고, 품사정보를 각 token의 추가한 sequence와, Naver DB를 통하여 얻은 named entity 정보를 입력으로 사용한다. 실험결과 cluster 방법으로 문제를 변형하고, sigmoid를 output layer의 activation function으로 사용하고 cross entropy cost function을 이용하여 network를 학습시켰을 때 F1 0.9873으로 가장 좋은 성능을 보였다.