• Proceedings of the Korean Society of Computer Information Conference
• /
• 2022.07a
• /
• pp.333-335
• /
• 2022

본 논문에서는 기존에 두 가지 모델이 필요했던 작업을 하나의 모델로 처리할 수 있는 자연어 처리 아키텍처를 제안한다. 단일 모델로 알츠하이머 환자의 언어패턴과 대화맥락을 분석하고 두 가지 결과인 환자분류와 챗봇의 대답을 도출한다. 일상생활에서 챗봇으로 환자의 언어특징을 파악한다면 의사는 조기진단을 위해 더 정밀한 진단과 치료를 계획할 수 있다. 제안된 모델은 전문가가 필요했던 질문지법을 대체하는 챗봇 개발에 활용된다. 모델이 수행하는 자연어 처리 작업은 두 가지이다. 첫 번째는 환자가 병을 가졌는지 여부를 확률로 표시하는 '자연어 분류'이고 두 번째는 환자의 대답에 대한 챗봇의 다음 '대답을 생성'하는 것이다. 전반부에서는 셀프어탠션 신경망을 통해 환자 발화 특징인 맥락벡터(context vector)를 추출한다. 이 맥락벡터와 챗봇(전문가, 진행자)의 질문을 함께 인코더에 입력해 질문자와 환자 사이 상호작용 특징을 담은 행렬을 얻는다. 벡터화된 행렬은 환자분류를 위한 확률값이 된다. 행렬을 챗봇(진행자)의 다음 대답과 함께 디코더에 입력해 다음 발화를 생성한다. 이 구조를 DementiaBank의 쿠키도둑묘사 말뭉치로 학습한 결과 인코더와 디코더의 손실함수 값이 유의미하게 줄어들며 수렴하는 양상을 확인할 수 있었다. 이는 알츠하이머병 환자의 발화 언어패턴을 포착하는 것이 향후 해당 병의 조기진단과 종단연구에 기여할 수 있음을 보여준다.

• Asia pacific journal of information systems
• /
• v.33 no.4
• /
• pp.1058-1092
• /
• 2023

This study investigates the perception and adoption of ChatGPT (a large language model (LLM)-based chatbot created by OpenAI) among Korean users and assesses its potential as the next disruptive innovation. Drawing on previous literature, the study proposes perceived intelligence and perceived anthropomorphism as key differentiating factors of ChatGPT from earlier AI-based chatbots. Four individual motives (i.e., perceived usefulness, ease of use, enjoyment, and trust) and two societal motives (social influence and AI anxiety) were identified as antecedents of ChatGPT acceptance. A survey was conducted within two Korean online communities related to artificial intelligence, the findings of which confirm that ChatGPT is being used for both utilitarian and hedonic purposes, and that perceived usefulness and enjoyment positively impact the behavioral intention to adopt the chatbot. However, unlike prior expectations, perceived ease-of-use was not shown to exert significant influence on behavioral intention. Moreover, trust was not found to be a significant influencer to behavioral intention, and while social influence played a substantial role in adoption intention and perceived usefulness, AI anxiety did not show a significant effect. The study confirmed that perceived intelligence and perceived anthropomorphism are constructs that influence the individual factors that influence behavioral intention to adopt and highlights the need for future research to deconstruct and explore the factors that make ChatGPT "enjoyable" and "easy to use" and to better understand its potential as a disruptive technology. Service developers and LLM providers are advised to design user-centric applications, focus on user-friendliness, acknowledge that building trust takes time, and recognize the role of social influence in adoption.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.34 no.3
• /
• pp.451-461
• /
• 2024

Recently, various AI chatbots based on large language models have been released. Chatbots have the advantage of providing users with quick and easy information through interactive prompts, making them useful in various fields such as question answering, writing, and programming. However, a vulnerability in chatbots called "prompt injection attacks" has been proposed. This attack involves injecting instructions into the chatbot to violate predefined guidelines. Such attacks can be critical as they may lead to the leakage of confidential information within large language models or trigger other malicious activities. However, the vulnerability of Korean prompts has not been adequately validated. Therefore, in this paper, we aim to generate malicious Korean prompts and perform attacks on the popular chatbot to analyze their feasibility. To achieve this, we propose a system that automatically generates malicious Korean prompts by analyzing existing prompt injection attacks. Specifically, we focus on generating malicious prompts that induce harmful expressions from large language models and validate their effectiveness in practice.

• Journal of Practical Engineering Education
• /
• v.16 no.1_spc
• /
• pp.33-39
• /
• 2024

This study examines the changes in learners' positive/negative perceptions of classroom experience and actual utilisation of AI chatbots in response to the recent changes in education trends caused by generative AI. AI chatbots were utilised in web programming classes for six classes of engineering students over two semesters. The learners' experience and usage were analysed from the beginning of the semester through surveys until the submission of midterm and final examination reports. The study's results indicate that the chatbot enhanced learning by providing Q/A feedback and solving practical problems. Additionally, the perception of the chatbot improved from midterm to the end of the course. The study also drew meaningful conclusions about the issue of community disconnection (personalisation) in the classroom and how to use it as educational software. This research is significant for the development of generative AI-based software.

• Journal of the korean academy of Pediatric Dentistry
• /
• v.51 no.3
• /
• pp.299-309
• /
• 2024

This study aimed to assess the competency of artificial intelligence (AI) in pediatric dentistry and compare it with that of dentists. We used open-source data obtained from the Korea Health Personnel Licensing Examination Institute. A total of 32 item multiple-choice pediatric dentistry exam questions were included. Two AI-based chatbots (ChatGPT 3.5 and Gemini) were evaluated. Each chatbot received the same questions seven times in separate chat sessions initiated on April 25, 2024. The accuracy was assessed by measuring the percentage of correct answers, and consistency was evaluated using Cronbach's alpha coefficient. Both ChatGPT 3.5 and Gemini demonstrated similar accuracy, with no significant differences observed between them. However, neither chatbot achieved the minimum passing score set by the Pediatric Dentistry National Examination. However, both chatbots exhibited acceptable consistency in their responses. Within the limits of this study, both AI-based chatbots did not sufficiently answer the pediatric dentistry exam questions. This finding suggests that pediatric dentists should be aware of the advantages and limitations of this new tool and effectively utilize it to promote patient health.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2017.04a
• /
• pp.958-959
• /
• 2017

본 논문은 챗봇과 대화시스템을 이용한 영어 교육 시스템을 기술하는 것을 목표로 한다. 본 논문의 시스템은 학습자의 대화 흐름을 제한하지 않고 주제를 벗어난 자유대화를 허용하며 문법오류에 대한 피드백을 한다. 챗봇과 대화시스템을 이용한 영어 교육 시스템은 대화턴 성공률로 평가되었는데, 평균 대화턴 성공률은 80.86%였으며, 주제별로는 1) 뉴욕시티투어 티켓 구매 71.86%, 2) 음식주문 71.06%, 3) 건강습관 대화 85.41%, 4) 미래화폐에 대한 생각 조사 95.09%였다. 또한 영어 문법 오류 교정도 측정되었는데 문법 오류 정확률은 66.7%, 재현율은 31.9%였다.

• Annual Conference on Human and Language Technology
• /
• 2018.10a
• /
• pp.35-39
• /
• 2018

챗봇은 사람과 기계가 자연어로 된 대화를 주고받는 시스템이다. 최근 대화형 인공지능 비서 시스템이 상용화되면서 일반적인 대화와 질의응답을 함께 처리해야할 필요성이 늘어나고 있다. 본 논문에서는 기계독해 기반 질의응답과 Transformer 기반 자연어 생성 모델을 함께 사용하여 하나의 모델에서 일반적인 대화와 질의응답을 함께 하는 기계독해 기반 질의응답 챗봇을 제안한다. 제안 모델은 기계독해 모델에 일반대화를 판단하는 옵션을 추가하여 기계독해를 하면서 자체적으로 문장을 분류하고, 기계독해 결과를 통해 자연어로 된 문장을 생성한다. 실험 결과 일반적인 대화 문장과 질의를 높은 성능으로 구별하면서 기계독해의 성능은 유지하였고 자연어 생성에서도 분류에 맞는 응답을 생성하였다.

• Knowledge Management Research
• /
• v.21 no.2
• /
• pp.177-196
• /
• 2020

Since the adoption of K-IFRS(Korean International Financial Reporting Standards), the amount of financial footnotes has been increased. However, due to the stereotypical phrase and the lack of conciseness, deriving the core information from footnotes is not really easy yet. To propose a solution for this problem, this study tried financial footnote analysis for financial ratio predictions based on text-mining techniques. Using the financial statements data from 2013 to 2018, we tried to predict the earning per share (EPS) of the following quarter. We found that measured prediction errors were significantly reduced when text-mined footnotes data were jointly used. We believe this result came from the fact that discretionary financial figures, which were hardly predicted with quantitative financial data, were more correlated with footnotes texts.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2018.07a
• /
• pp.183-184
• /
• 2018

우리나라의 노인자살률은 OECD 국가 중에 1위이다. 인위적 고독사인 '자살'의 가장 큰 원인인 우울증을 의사소통을 통해 예방하고자 한다. 본 논문에서는 상황에 따라 독거노인에게 스피커가 먼저 질문하는 형식인 시스템을 제안한다. 음성인식 시스템인 스피커를 활용하여 독거노인의 의사소통을 증대시키고, 질문뿐만이 아니라 식사 여부, 약 복용 여부 관련 일상 알람도 주기 때문에 규칙적인 생활을 하는 데 도움을 준다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2017.11a
• /
• pp.671-674
• /
• 2017

고객 상담 채팅은 비동기 상태에서도 메시지를 보낼 수 있기 때문에 고객이 메시지를 보낸 후 한참 후에 메시지를 읽고 답을 보내는 경우가 많아 하나의 고객 문의를 처리하는데 시간이 많이 소요된다. 이런 문제점을 해결하기 위해서 본 논문에서는 룰 엔진 기반의 챗봇 IFTTT 서비스 시스템을 제안한다. 이를 통하여 고객 상담 업무를 자동적으로 실시간 처리 할 수 있다.