• The KIPS Transactions:PartD
• /
• v.11D no.5
• /
• pp.1177-1188
• /
• 2004

Multi-agent systems have been developed for supporting intelligent collaboration of distributed and independent software entities and are be-ing widely used for various applications. For the collaboration among agents, conversation policies (or interaction protocols) mutually agreed by agents are used. In today's dynamic electronic market environment, there can be frequent changes in conversation policies induced by the changes in transaction methods in the market, and thus, the importance of ad hoc conversation policies is increasing. In existing agent platforms, they allow the use of only several standard or fixed conversation policies, which requires inevitable re implementation for ad hoc conversation policies and leads to inefficiency and intricacy. This paper designs an agent platform that supports ad hoc conversation policies and presents the prototype implementation. The suggested system includes an exchangeable and interpretable conversation policy model, a meta conversation procedure for exchanging new conversation policies, and a mechanism for performing actual transactions with exchanged conversation policies in run time in an adaptive way.

• Journal of the Korea Society of Computer and Information
• /
• v.13 no.7
• /
• pp.59-66
• /
• 2008

This paper describes a novel method for identifying the main topic and detecting topic changes in a text-based dialogue as in Instant Messaging (IM). Compared to other forms of text, dialogues are uniquely characterized with the short length of text with small number of words, two or more participants, and existence of a history that affects the current utterance. Noting the characteristics, our method detects the main topic of a dialogue by considering the keywords not only the utterances of the user but also the dialogue system's responses. Dialogue histories are also considered in the detection process to increase accuracy. For topic change detection, the similarity between the former utterance's topic and the current utterance's topic is calculated. If the similarity is smaller than a certain threshold, our system judges that the topic has been changed from the current utterance. We obtained 88.2% and 87.4% accuracy in topic detection and topic change detection, respectively.

• Annual Conference on Human and Language Technology
• /
• 2020.10a
• /
• pp.491-494
• /
• 2020

최근 언어 모델(Language model)의 기술이 발전함에 따라, 자연어처리 분야의 많은 연구들이 좋은 성능을 내고 있다. 정해진 주제 없이 인간과 잡담을 나눌 수 있는 오픈 도메인 대화 시스템(Open-domain dialogue system) 분야에서 역시 이전보다 더 자연스러운 발화를 생성할 수 있게 되었다. 언어 모델의 발전은 응답 선택(Response selection) 분야에서도 모델이 맥락에 알맞은 답변을 선택하도록 하는 데 기여를 했다. 하지만, 대화 모델이 답변을 생성할 때 일관성 없는 답변을 만들거나, 구체적이지 않고 일반적인 답변만을 하는 문제가 대두되었다. 이를 해결하기 위하여 화자의 개인화된 정보에 기반한 대화인 페르소나(Persona) 대화 데이터 및 태스크가 연구되고 있다. 페르소나 대화 태스크에서는 화자마다 주어진 페르소나가 있고, 대화를 할 때 주어진 페르소나와 일관성이 있는 답변을 선택하거나 생성해야 한다. 이에 우리는 대용량의 코퍼스(Corpus)에 사전 학습(Pre-trained) 된 언어 모델을 활용하여 더 적절한 답변을 선택하는 페르소나 대화 시스템에 대하여 논의한다. 언어 모델 중 자기 회귀(Auto-regressive) 방식으로 모델링을 하는 GPT-2, DialoGPT와 오토인코더(Auto-encoder)를 이용한 BERT, 두 모델이 결합되어 있는 구조인 BART가 실험에 활용되었다. 이와 같이 본 논문에서는 여러 종류의 언어 모델을 페르소나 대화 태스크에 대해 비교 실험을 진행했고, 그 결과 Hits@1 점수에서 BERT가 가장 우수한 성능을 보이는 것을 확인할 수 있었다.

• Smart Media Journal
• /
• v.12 no.11
• /
• pp.77-85
• /
• 2023

After COVID-19, communication through online platforms has increased, leading to an accumulation of massive amounts of conversational text data. With the growing importance of summarizing this text data to extract meaningful information, there has been active research on deep learning-based abstractive summarization. However, conversational data, compared to structured texts like news articles, often contains missing or transformed information, necessitating consideration from multiple perspectives due to its unique characteristics. In particular, vocabulary omissions and unrelated expressions in the conversation can hinder effective summarization. Therefore, in this study, we restructured by considering the characteristics of Korean conversational data, fine-tuning a pre-trained text summarization model based on KoBART, and improved conversation data summary perfomance through a refining operation to remove redundant elements from the summary. By restructuring the sentences based on the order of utterances and extracting a central speaker, we combined methods to restructure the conversation around them. As a result, there was about a 4 point improvement in the Rouge-1 score. This study has demonstrated the significance of our conversation restructuring approach, which considers the characteristics of dialogue, in enhancing Korean conversation summarization performance.

• Annual Conference on Human and Language Technology
• /
• 2017.10a
• /
• pp.111-115
• /
• 2017

목적 지향적 대화 시스템(Goal-oriented dialogue system) 은 텍스트나 음성을 통해 특정한 목적을 수행 할 수 있는 시스템이다. 최근 RNN(recurrent neural networks)을 기반으로 대화 데이터를 end-to-end learning 방식으로 학습하여 대화 시스템을 구축하는데에 활용한 연구가 있다. End-to-end 방식의 학습은 도메인에 대한 지식 없이 학습 데이터 자체만으로 대화 시스템 구축을 위한 학습이 가능하다는 장점이 있지만 도메인 지식을 학습하기 위해서는 많은 양의 데이터가 필요하다는 단점이 존재한다. 이에 본 논문에서는 도메인 특정 지식을 결합하여 end-to-end learning 방식의 학습이 가능한 Hybrid Code Network 구조를 기반으로 한국어로 구성된 식당 예약에 관련한 대화 데이터셋을 이용하여 식당 예약을 목적으로하는 대화 시스템을 구축하는 방법을 제안한다. 실험 결과 본 시스템은 응답 별 정확도 95%와 대화 별 정확도 63%의 성능을 나타냈다.

• Annual Conference on Human and Language Technology
• /
• 1997.10a
• /
• pp.271-276
• /
• 1997

대화체는 문어체와는 달리 화자와 청자 사이의 질의/응답으로 이루어진 형태의 문장들을 가지며, 생략과 대용어가 빈번히 발생하는 특징을 갖는다. 이러한 대화 형태에서 어떠한 한 문장에는 화자가 전달하고자 하는 의도를 포함하고 있다. 이러한 대화체 문장들을 번역하는 것은 단순한 언어적 분석에 의한 번역으로서는 많은 번역상의 오류가 발생하게 된다. 따라서 대화체 문장들의 올바른 번역을 위해서는 대화의 상황을 반영하는 문맥 정보가 부가적으로 요구된다. 본 연구에서는 이러한 문맥 정보로서 화행을 사용하여 대화체 기계번역을 수행하고자 한다. 화행(Speech Act)이란 화자에 의해 의도되어 발화 속에 포함된 언어적 행위를 나타내며, 이러한 화행을 분석함으로써 화자의 의도를 파악하고 이를 통해 올바른 번역을 수행할 수 있게 된다. 본 기계번역 시스템에 포함된 화행 분석 과정에서는 대화를 화행으로 모델링한 담화 문법과 유사한 형태의 재귀적 대화 전이망(Recursive Dialog Transition Network)을 사용하게 된다. 본 논문에서는 호텔 예약 영역에서의 기계번역 시스템에 대한 간단한 소개와 화행의 종류 및 분석 방법과 이를 통한 기계번역 방식에 대해 살펴보도록 하겠다.

• The Journal of the Acoustical Society of Korea
• /
• v.21 no.3
• /
• pp.330-338
• /
• 2002

Spontaneous speech is ungrammatical as well as serious phonological variations, which make recognition extremely difficult, compared with read speech. In this paper, for conversational speech recognition, we analyze the transcriptions of the real conversational speech, and then classify the characteristics of conversational speech in the speech recognition aspect. Reflecting these features, we obtain the baseline system for conversational speech recognition. The classification consists of long duration of silence, disfluencies and phonological variations; each of them is classified with similar features. To deal with these characteristics, first, we update silence model and append a filled pause model, a garbage model; second, we append multiple phonetic transcriptions to lexicon for most frequent phonological variations. In our experiments, our baseline morpheme error rate (WER) is 31.65%; we obtain MER reductions such as 2.08% for silence and garbage model, 0.73% for filled pause model, and 0.73% for phonological variations. Finally, we obtain 27.92% MER for conversational speech recognition, which will be used as a baseline for further study.

• 재활복지
• /
• v.22 no.4
• /
• pp.103-121
• /
• 2018

For knowing the effects of aging on conversational skills in daily communication, this paper studied for the conversational turn-taking skills, working memory and naming ability on healthy elderly adults over 65 ages. 85 elderly adults participated in this study, which divided into four groups by ages. Speech samples were collected in natural conversation. Memorization of numbers, mental calculation, repetition of words were administered for working memory test. K-BNT was used for the naming ability. One-way ANOVA analysis was used for the comparison of conversational turn-taking skills among four groups. We analyzed the correlation between conversational skills, working memory and naming ability. The results were as follows: first, there were a significant difference in conversational turn-taking skills by age, but not by gender. There was a significant difference in 'Turn-Taking Frequency' and 'Total Utterance Frequency' among four groups. The same results were shown in the scores of females within three groups(exclude groups over 85D)(p<.01). Second, there was a significant correlation between 'rates of maintenance' and 'naming ability'. In addition, it was found that the naming test predicted 'rates of maintenance' skills. The results of this study suggest that word-retrieval ability will be helpful to enhance functional communication skills in aging old adults.

• Korean Journal of Cognitive Science
• /
• v.19 no.2
• /
• pp.107-123
• /
• 2008

The existing studies of a dialogue system can be classified into two major parts. One is a study for a practical system, and the other is a study to understand a principal of dialogue phenomena. The former focuses on robustness in real environment for dialogue systems. However, it cannot guarantee its performance in complicated dialogue environment. The latter has studied as the plan-based model typically. It has strong points that it can reflect complex dialogue phenomena and can infer user's intention in various situations. However, an initial design of this model is so complicated, and it is difficult for this model to be extended to the interaction model for response generation in a practical dialogue system. This paper proposes a new dialogue modeling using plan recognition and a discourse stark to effectively generate response in a practical dialogue system.

• Archives of design research
• /
• v.16 no.1
• /
• pp.117-126
• /
• 2003

The human computer interaction has gone through diverse changes along with the technical progress of computers. With the spread of the world wide web, attempts of various forms and methods of interactions reflecting the characteristics of the web are being made. Among the diverse concepts of interaction, Conversational Agent uses a character as the visual perceptional object and through the interaction based on conversation it suggests a natural, human, intuitive interaction. In this study, Conversational Agent was suggested and applied on a cyber lecture website and the appropriateness was verified. The fundamental structure and method needed in constructing a conversation was examined through the study on conversation. After that the characteristics of the conversation used on the web was identified and hereby and the coherence and conversation structure was deducted. After the characteristics of the conversational agent were observed, it was applied to the previously deducted web sites logical connection and conversation structure. Lastly, the conversational agent suitable for the web site was proposed and realized.