Text-Driven Multiple-Path Discourse Processing for Descriptive Texts

Jungyun Seo 대한전기학회 1996 Journal of Electical Engineering and information S Vol.1 No.2

Text-Driven Multiple-Path Discourse Processing for Descriptive Texts

Jungyun Seo 한국정보과학회 1996 Journal of Electrical Engineering and Information Vol.1 No.2

This paper presents a text-driven discourse analysis system, called DPAS. DPAS constructs a discourse structure by weaving together clauses in the text by finding discourse relations between a clause and the clauses in a context. The basic processing model of DPAS is based on the stack based model of discourse analysis suggested by Grosz and Sidner. We extend the model with dynamic programming method to handle various discourse ambiguities effectively and efficiently. We develop the idea of a context space to keep all information of a context. DPAS parses a text by considering all possible discourse relations between a clause and a context. Since different discourse relations may result in different states of a context, DPAS maintains multiple context spaces for an ambiguous text. Since maintaining all interpretations until the whole text is processed requires too much computing resources, DPAS uses the idea of depth-limited search to limit the search space. If there is more than one discourse relation between an input clause and a context, DPAS constructs context spaces―one context space for each discourse relation. Then, DPAS applies heuristics to choose the most desirable context space after it processes some more input clauses. Since the basic idea of DPAS is domain independent, although we used descriptive texts to demonstrate DPAS, we believe the idca of DPAS can be extended to understand other styles of texts.

Text-Driven Multiple-Path Discourse Processing for Descriptive Texts

Seo, Jungyun The Korean Institute of Electrical Engineers 1996 Journal of Electrical Engineering and Information Vol.1 No.2

This paper presents a text-driven discourse analysis system, called DPAS. DPAS constructs a discourse structure by weaving together clauses in the text by finding discourse relations between a clause and the clauses in a context. The basic processing model of DPAS is based on the stack based model of discourse analysis suggested by Grosz and Sidner. We extend the model with dynamic programming method to handle various discourse ambiguities effectively and efficiently. We develop the idea of a context space to keep all information of a context. DPAS parses a text by considering all possible discourse relations between a clause and a context. Since different discourse relations may result in different states of a context, DPAS maintains multiple context spaces for an ambiguous text. Since maintaining all interpretations until the whole text is processed requires too much computing resources, DPAS uses the idea of depth-limited search to limit the search space. If there is more than one discourse relation between an input clause and a context, DPAS constructs context spaces one context space for each discourse relation. Then, DPAS applies heuristics to choose the most desirable context space after it processes some more input clauses. Since the basic idea of DPAS is domain independent, although we used descriptive texts to demonstrate DPAS, we believe the idea of DPAS can be extended to understand other styles of texts.

Journal of Computing Science and Engineering(JCSE)의 국제화 작업

서정연(Jungyun Seo) 한국정보과학회 2010 정보과학회지 Vol.28 No.10

방출셀을 이용한 액상건축자재 오염물질 방출시험방법 정립에 관한 연구

임정연(Jungyun Lim),장성기(Seongki Jang),서수연(Sooyun Seo) 한국분석과학회 2009 분석과학 Vol.22 No.3

건물 경계 그래프를 이용한 NLOS GPS 신호 탐지

변정윤(Jungyun Byun),김상현(Sanghyun Kim),서지원(Jiwon Seo) 한국통신학회 2022 한국통신학회 학술대회논문집 Vol.2022 No.2

본 논문에서는 NLOS(non-line-of-sight) GPS 신호 탐지를 위해 3D 도시 모델을 기반으로 건물경계 그래프를 제작하고, 실제 GPS 신호를 바탕으로 탐지 성능을 검증하였다. 그 결과, 본 논문에서 제작한 건물 경계 그래프가 NLOS 신호를 탐지하는데 적절히 사용될 수 있음을 보였다.

Issues and Empirical Results for Improving Text Classification

Youngjoong Ko,Jungyun Seo 한국정보과학회 2011 Journal of Computing Science and Engineering Vol.5 No.2

Automatic text classification has a long history and many studies have been conducted in this field. In particular, many machine learning algorithms and information retrieval techniques have been applied to text classification tasks. Even though much technical progress has been made in text classification, there is still room for improvement in text classification. In this paper, we will discuss remaining issues in improving text classification. In this paper, three improvement issues are presented including automatic training data generation, noisy data treatment and term weighting and indexing, and four actual studies and their empirical results for those issues are introduced. First, the semi-supervised learning technique is applied to text classification to efficiently create training data. For effective noisy data treatment, a noisy data reduction method and a robust text classifier from noisy data are developed as a solution. Finally, the term weighting and indexing technique is revised by reflecting the importance of sentences into term weight calculation using summarization techniques.

Improving Educational Dialog System for Historical World Affairs

Eunkyoung Jo,Jungyun Seo 한국응용언어학회 2021 응용 언어학 Vol.37 No.Special

This study aims to introduce the process of developing a dialog system to inform users of historical world affairs. It further aims to show the process of improving the dialog system by expanding the users’ question phrases to expose the contents maximally. The dialog system’s contents contain testimonial talks made by Korean comfort women survivors of Japanese army sexual slavery during World War II. We built a dialog system to design interactions between a user and video-recorded testimonies. In the dialog system, a user is able to interact with them verbally and virtually. The system plays the role of the survivors, who underwent painful experiences during the period. In this dialog system—of which knowledge content for testimonial answers is limited—it is more important to maximally cover the prepared answers than emulate dynamic conversation, which many dialog application researchers have focused on so far to let dialogs have sequentially diverse. Therefore, we expanded question phrases using lexical relatedness and various modal expressions, thereby boosting the content coverage. Overall, this study aims to seek a digital way of representing a dialog on historical world affairs with limited contents.

A Statistical Learning Method for Identification of Analysis Classes from Requirements in Korean

Hyoungil Jeong,Jungyun Seo,Soojin Park 보안공학연구지원센터 2013 International Journal of Multimedia and Ubiquitous Vol.8 No.1

다영역 대화 모델링을 위한 다중 맥락 기법

김지훈(Jihoon Kim),서정연(Jungyun Seo) 한국인지과학회 1998 인지과학 Vol.9 No.3

자연언어를 이용한 대화 인터페이스의 기본 목적은 자연스러운 인간의 대화를 통한 인간과 컴퓨터간의 의사소통이다. 인간과 대화를 진행하게 되는 시스템은 사용자 발화의 의미를 파악해 야함은 물론이고 사용자의 의도, 대화 목적 등을 파악할 수 있어야 한다. 대화시스템은 대화모델을 통해 대화를 자연스럽게 진행해 갈 수 있고 적절한 시스템 발화를 결정할 수 있으며, 나아가 사용자의 다음 발화를 예측 가능하게 된다. 그러나, 상호 독립적인 여러 개의 영역으로 구성되어 있는 다영역 대화 시스템의 경우 단일 영역 대화와는 달리 영역 상호간 참조를 위한 대화의 중단과 재기현상이 빈번히 발생한다. 특히 음성 인식과 통합되어 실시간으로 운용되게 되는 다영역 자연언어 대화 인터페이스의 대화모델은 안정성, 효율성과 더불어, 추가로 다중 영역 대화의 병렬 대화 현상의 제어기능을 갖추어야 한다는 것이다. 본 논문에서는 실시간 음성 인식을 이용한 자연어 인터페이스에 적합한 다영역 대화모델을 위해 대화흐름도를 이용한 대화 진행 기법을 제안하고. 다중 영역 대화에서 빈번하게 발생하는 대화의 중단, 재기 현상을 제어할 수 있는 다중 대화 맥락의 유지가 가능한 대화모델을 제안한다. Natural language dialogue interface techniques allow users to communicate with computers using human languages. A dialogue interface system should not only understand users utterances but also detect the users intentions. A dialogue interface system must have proper strategies for dialogues, called a dialogue model, for target domains. With a dialogue model, it can control dialogues in natural manner, decide the next utterances, and, most importantly, predict the users following utterances. A dialogue model usually strongly tied with domain knowledge of a target domain. In a multi-domain dialogue interface system, however, a user can make utterances in several domains in one dialogue. During a dialogue, furthermore, the user makes an interrupt to get some information in other domains and resumes dialogue with the information. To make a real-time speech dialogue interface system reliable and efficient, it is important to design a dialogue model so that it can control over the parallel dialogue phenomena frequently occurred in multi-domain dialogues. In this paper, we suggest a processing model for natural language interface using real-time speech recognition. We also propose a new dialogue model that maintains multi-dialogue threads that can solve dialogue interrupts and resume phenomena we often see in multi-domain dialogues.