• Journal of the Korean Data Analysis Society
• /
• v.20 no.6
• /
• pp.2883-2893
• /
• 2018

We suggest a stratified conditional unrelated question randomized response model to more efficiently estimate a sensitive character A when the population is composed of several strata. In that model, only the respondents who answered "yes" through randomization device which was consisted of a less sensitive character B and a question forcing to answer "yes" respond to our suggested model and we deal with two allocation problems of proportional allocation and optimal one. We expand the suggested model into two sample stratified conditional unrelated question model to cover the case of unknowing unrelated character and deduce minimal variance through optimal sample size of stratum h. Finally, we show that the suggested model is more efficiency than stratified unrelated models and the stratified Carr et al.'s model (1982) under some given conditions, and show numerically that the smaller the values ${\pi}_{h2}$ and ${\pi}_{hy}$, the more efficiency the fit of the model.

• Proceedings of the Korean Information Science Society Conference
• /
• 2005.07b
• /
• pp.511-513
• /
• 2005

본 논문은 정답 색인 방법을 이용하여 응답 속도가 빠르고 정확한 백과사전 질의응답 시스템을 구현하는 방법을 제안한다. 논문에서 제안한 정답 색인 방법은 대상 문서에서 160여 개의 정답 유형 범주에 해당하는 정답 후보를 인식하고, 정답 후보와 색인 범주에 속하는 키워드를 색인단위로 정의하여 저장하였다. 특히 용언정보에 대해서는 LF(Logical Form)단위로 색인하여 색인 정확도를 높였다. 정답 랭킹에서는 사용자 질문에서 각 단어별로 문장 성분. 단어 가중치 정보 등을 이용하여, 필수단어를 산정하고 이를 정답랭킹의 방법으로 활용하였다. 이러한 방법론은 용언 정보를 활용해야 효과적인 백과사전이라는 문서 도메인의 특성을 반영하고, 빠른 질문 응답 시간을 보장하는 백과사전 질의응답 시스템에 적합하다.

• Survey Research
• /
• v.18 no.4
• /
• pp.127-147
• /
• 2017

Despite the importance of questionnaires, little survey methodology research on questionnaire design has been conducted in Korea. The purpose of this study was to explore whether two questionnaire forms (close-ended vs. open-ended questions) about 'the most important problem in Korea' elicited similar responses. During the 2016 Korean General Social Survey (KGSS), a random half of respondents were asked the open-ended question form and the remaining half were asked the close-ended question form. While the economy is the most mentioned response (35% vs. 33.2%) to both close-ended and open-ended question forms, there is similarity in the order of highly mentioned responses if we consider that 'politics' is not provided as one of response categories in the close-ended question form. The order of second to fourth response category is crime (24.4%), education (15.4%), and poverty (6.3%) to the closed-ended question form, and politics (10.8%), crime (9.5%), and education (7.6%) to the open-ended question form. Also, the characteristics of respondents who responded with the economy as being the most important are slightly different between the two halves in terms of age, household income, and satisfaction with economic condition. Our findings suggest that we need to be careful when we adopt questions developed in other countries and to consider using survey experiments in pre-testing questionnaire items.

• Journal of KIISE
• /
• v.43 no.8
• /
• pp.902-909
• /
• 2016

Question answering system (QA system) is a system that finds an actual answer to the question posed by a user, whereas a typical search engine would only find the links to the relevant documents. Recent works related to the open domain QA systems are receiving much attention in the fields of natural language processing, artificial intelligence, and data mining. However, the prior works on QA systems simply replace all words that are not in the training data with a single token, even though such unseen words are likely to play crucial roles in differentiating the candidate answers from the actual answers. In this paper, we propose a method to compute vectors of such unseen words by taking into account the context in which the words have occurred. Next, we also propose a model which utilizes inverse document frequencies (IDF) to efficiently process unseen words by expanding the system's vocabulary. Finally, we validate that the proposed method and model improve the performance of a QA system through experiments.

• Journal of Korea Society of Industrial Information Systems
• /
• v.28 no.5
• /
• pp.31-39
• /
• 2023

The evolution of Natural Language Processing (NLP) and the rise of large language models (LLM) like ChatGPT have paved the way for specialized question-answering (QA) systems tailored to specific domains. This study outlines a system harnessing the power of LLM in conjunction with document search algorithms to interpret and address user inquiries using documents from the Korea Atomic Energy Research Institute (KAERI). Initially, the system refines multiple documents for optimized search and analysis, breaking the content into managable paragraphs suitable for the language model's processing. Each paragraph's content is converted into a vector via an embedding model and archived in a database. Upon receiving a user query, the system matches the extracted vectors from the question with the stored vectors, pinpointing the most pertinent content. The chosen paragraphs, combined with the user's query, are then processed by the language generation model to formulate a response. Tests encompassing a spectrum of questions verified the system's proficiency in discerning question intent, understanding diverse documents, and delivering rapid and precise answers.

• 전기의세계
• /
• v.44 no.6
• /
• pp.40-45
• /
• 1995

3일간의 심포지움을 통하여 느낀점을 몇가지 적어보면 다음과 같다. 참서자들이 첫날부터 마지막 날까지 오전, 오후 1회의 coffee break를 제외하고는 하루 종일 강연회장에 앉아서 열심히 경청하며 노트하고, 질문하는 등 열의가 대단했다. 또한 일본인 발표자 및 질문자를 위해 동시 통역을 고용하였는데, 영어 소통능력이 부족하더라도 당당하게 일어로 질의 응답을 함으로써 정보 교류가 가능하도록 배려한 점은 앞으로 우리나라에서 이와 같은 심포지움을 개최하게 될때 검토해야 할 것 같다. 특히 통역이 여성이었는데 전력 반도체 관련 용어에 매우 정통해 있고, 사전에 논문을 검토하여 질의 응답시 전문가 이상으로 능력을 발휘하는 것을 보고 우리도 이와 같은 전문 분야별 통역의 양성이 필요하다는 것을 느꼈다.

• Environmental and Resource Economics Review
• /
• v.13 no.3
• /
• pp.387-415
• /
• 2004

This paper is intended to valuate air quality of the Seoul Metropolitan Area using triple-bound dichotomous choice (TBDC) contingent valuation method (CVM), supplemented by open-ended (OE) questionnaires. In the OE questionnaires, some respondents would state their willingness to pay (WTP) outside the limits of the WTP interval. It implies that WTP estimates based on the customary dichotomous choice (DC) questionnaires can be biased. We argue that the TBDC-CVM refined with OE questions is more efficient, because the latter helps purge the former of corrupted data that may have been collected by the TBDC interview process.

• Annual Conference on Human and Language Technology
• /
• 2019.10a
• /
• pp.315-320
• /
• 2019

질문 생성이란 구절이 입력되면 구절에서 답을 찾을 수 있는 질문을 자동으로 생성하는 작업으로 교육용 시스템, 대화 시스템, QA 시스템 등 다양한 분야에서 중요한 역할을 한다. 질문 생성에서 정답의 단어가 질문에 포함되는 문제점을 해결하기 위해 구절과 정답을 분리한다. 하지만 구절과 정답을 분리하게 되면 구절에서 정답의 정보가 손실되고, 정답에서는 구절의 문맥 정보가 손실되어 정답 유형에 맞는 질문을 생성할 수 없는 문제가 발생된다. 본 논문은 이러한 문제를 해결하기 위해 분리된 정답과 구절의 정보를 연결시켜주는 정답과 구절의 공동 주의 집중 계층을 제안한다. 23,658개의 질문-응답 쌍의 말뭉치를 이용한 실험에서 정답과 구절의 공동 주의 집중 계층이 성능 향상에 기여해 우수한 성능(BLEU-26.7, ROUGE-57.5)을 보였다.

• The Korean Journal of Applied Statistics
• /
• v.17 no.2
• /
• pp.303-310
• /
• 2004

We suggest a method to procure information from the sensitive population which combine a direct survey method, BB and an indirect survey one, RRT, and a combined estimator that uses the stratified double sampling to estimate the sensitive parameter. We compare the efficiency of our estimator with that of Mangat and Singh model.

• Annual Conference on Human and Language Technology
• /
• 2022.10a
• /
• pp.95-99
• /
• 2022

오픈 도메인 질의 응답 (ODQA, Open-Domain Question Answering)은 주어진 질문에 대한 답을 찾는 과업으로서 질문과 관련있는 지식을 찾는 "검색" 단계를 필요로 한다. 최근 이미지, 테이블 등의 검색을 요구하는 멀티 모달 ODQA에 대한 연구가 많이 진행되었을 뿐만 아니라 산업에서의 중요도 또한 높아지고 있다. 본 논문은 여러 종류의 멀티 모달 ODQA 중에서도 테이블 - 텍스트 기반 멀티 모달 ODQA 데이터 집합으로 Fusion-in-Decoder (FiD)를 이용한 멀티 모달 오픈 도메인 질의 응답 연구를 제안하며 베이스라인 대비 최대 EM 20.5, F1 23.2 향상을 보였다.