• Annual Conference on Human and Language Technology
• /
• 2018.10a
• /
• pp.665-668
• /
• 2018

본 논문에서는 제약기반 KBQA를 위한 질문분석 기술에 대해서 소개한다. 핵심개체와 속성에 대한 연결 모호성을 해소하기 위해서 세 종류의 제약정보 활용을 제안한다. 세 종류의 제약은 핵심개체에 기반한 제약, 의미정답유형에 기반한 제약, 속성단서에 기반한 제약이다. 제약을 위해서는 질문 내에서 핵심개체와 속성단서를 인식하여야 한다. 본 논문에서는 규칙과 휴리스틱에 기반한 핵심개체와 속성단서 인식 방법에 대해서 소개한다. 핵심개체와 속성단서 인식 실험은 구축된 229개의 질문을 대상으로 수행하였으며, 핵심개체와 속성단서가 모두 정확히 인식된 정확도(accuracy)가 57.21%이고, KBQA 대상질문에서는 71.08%를 보였다.

• The Journal of the Korea Contents Association
• /
• v.22 no.3
• /
• pp.134-140
• /
• 2022

The purpose of this study was to investigate the directives and Wh-words in the directives from elementary 2nd, 4th and 6th grade Korean textbooks. After entering all directives into Microsoft Office Excel, directives with Wh-words were separated. The analysis program, Natmal, was used for the analysis of the directives and Wh-words. The criteria from previous studies were also applied for this analysis process. As a result of the study, there are a lot of nouns and verbs in directives. They were consisted of sentences with an average of 6.9 Eojeol. There were a total of 11 types of Wh-words and 'Mueot(what), Eotteon(which), eotteohge(how)' appeared most frequently in all grades. For question types, both grades had more inferential questions than literal information questions. This results were expected to be used as basic data for language interventions with school aged children who have language disorders.

• Journal of The Korean Association For Science Education
• /
• v.30 no.6
• /
• pp.799-814
• /
• 2010

This study aimed at developing a set of question cards for fostering deep understanding and encouraging reasoning about fossils and analyze the characteristics of visitors' communication depending on whether to use the question cards in a fossil gallery. Through several steps, a card set consisted of nine generic questions about fossil exhibitions and guidance for using question cards were developed. Data related to visitors' communications were collected from 18 peer groups (from 5th to 9th grade) visiting the fossil gallery of Gwacheon National Science Museum. Visiting groups' interactions were videotape recorded and transcribed. 'Holding time,' the types of 'actions,' and the types of 'conversation' were analysed. Visitors' actions were divided into three categories: ‘look’, 'speech', and 'motion.' Furthermore, visitors' conversations categorized as 'speech' were subdivided into four patterns: 'enumerative,' 'consensual,' 'responsive,' and 'argumentative.' Using the question cards contributes to increase holding time and most of the visiting actions. Most of the conversation patterns also increased except the responsive pattern. In conclusion, using question cards in a fossil gallery could facilitate concentrated and meaningful visits by enhancing active verbal and non-verbal communications between exhibit and visitor or among visitors, encouraging visitors' reasoning about exhibits, and guiding visitors what and how to focus on exhibits.

• Journal of Information Management
• /
• v.37 no.2
• /
• pp.137-156
• /
• 2006

To evaluate the performance of the question and answer services provided through the internet portals in Korea, question and answer transcript of four major services were sampled systematically. Using the digital reference evaluation framework, number and types of questions, response rate, timeliness, accuracy for information questions and user satisfaction were measured and analyzed. The level of the service performance is identified and compared. The conclusion includes suggestions for service improvement.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2020.11a
• /
• pp.984-987
• /
• 2020

최근 오픈 도메인 자연어 질문 응답 분야에서는 폭넓은 다중 문서들을 토대로 다중 홉 추론과 동시에 서로 다른 수준의 여러 문제들을 한꺼번에 해결해야 하는 다중 작업 질문 응답에 관한 관심이 높다. 본 논문에서는 이러한 다중 홉 추론과 다중 작업을 요구하는 복잡 질문들에 효과적으로 응답하기 위해, 계층적 그래프 기반의 새로운 심층 신경망 모델을 제안한다. 제안 모델에서는 계층적 그래프와 그래프 신경망을 이용해 다중 문서들로부터 서로 다른 수준의 맥락 정보를 얻어낸 후, 이들을 활용하여 뒷받침 문장들, 답변 영역, 응답 유형 등을 동시에 구해야 하는 다중 작업 문제에 관한 답들을 예측해낸다. 본 논문에서는 오픈 도메인 자연어 질문 응답 데이터 집합인 HotpotQA를 이용한 실험들을 통해, 제안 모델의 긍정적 효과를 입증한다.

• Journal of Korean Library and Information Science Society
• /
• v.38 no.3
• /
• pp.223-246
• /
• 2007

This study examines the retrieval efficiency of the FRBR-based bibliographic retrieval system. To do this, we built two experimental retrieval systems(a FRBR-based system constructed through FRBRizing algorithms and an OPAC-based retrieval system) using 387 music materials coded in a KORMARC format. Next, we set up six hypotheses and compared these two systems in terms of recall, precision, and retrieval time using 28 participants and a questionnaire with 12 queries. The results show that the average recall value of the FRBR-based system Is higher than that of the OPAC system regardless of query types and the average precision and retrieval time values of manifestation queries of the OPAC system is more efficient that those of the FRBR-based system. This study results can be used to customize digital library interfaces as well as to improve the retrieval efficiency of the bibliographic retrieval system.

• Journal of the Korean Chemical Society
• /
• v.55 no.1
• /
• pp.124-131
• /
• 2011

This study is conducted to analyze the questioning styles of the parts of chemistry in 17 middle school science textbooks in terms of frequency, types, and placement of questions. The instrument was the Textbook Questioning Strategies Assessment Instrument (TQSAI) developed by the Cooperative Teacher Preparation Program, University of California in 1975. An averaged ratio of questionings/sentences (Q/S) of seventeen textbooks was 13.6% varying from 6.8% to 19.7%. Various question types were used in titles, early and middle steps in the sections except late steps. The comprehending questions were most frequent in the types of inquiry courses and the focusing questions were most frequent in the characteristics of instructional courses. The percentages of empirical questions for the total numbers of questions were slightly larger than nonempirical questions with values of 54.2% and 45.8%, respectively.

• Journal of The Korean Association For Science Education
• /
• v.38 no.2
• /
• pp.235-247
• /
• 2018

The student's pre-questions (pre-class questions related to the learning contents) not only provide the teacher a gauge of the interest and level of the student, but also provide a useful means of providing clues to proceed with the teaching-learning process. The purpose of this study is to develop an analytical framework for effectively analyzing students' pre-questions and to analyze students' pre-questions related to elementary science learning unit of the 2009 revised curriculum by applying this framework. The developed framework is composed of three major categories: knowledge type, extended type, and curious type, each of which is then subdivided into several sub-categories. Using the developed analysis framework, 914 pre-questions from the students presented in the $5^{th}$ and $6^{th}$ grades of elementary science in the 2009 revised curriculum were analyzed, and the types of questions distributed by grade. The percentage of questions by type was also different. Based on the results of this study, students' needs for learning can be grasped through the pre-questions analysis framework and reflected in the teaching-learning process, and student-centered learning contents and methods could be presented. It is expected to make a meaningful contribution to the analysis framework.

• KIPS Transactions on Software and Data Engineering
• /
• v.4 no.7
• /
• pp.283-290
• /
• 2015

In this paper, we propose a hybrid method for the recognition of answer types in the WiseQA system. The answer types are classified into two categories: the lexical answer type (LAT) and the semantic answer type (SAT). This paper proposes two models for the LAT detection. One is a rule-based model using question focuses. The other is a machine learning model based on sequence labeling. We also propose two models for the SAT classification. They are a machine learning model based on multiclass classification and a filtering-rule model based on the lexical answer type. The performance of the LAT detection and the SAT classification shows F1-score of 82.47% and precision of 77.13%, respectively. Compared with IBM Watson for the performance of the LAT, the precision is 1.0% lower and the recall is 7.4% higher.

• KIPS Transactions on Software and Data Engineering
• /
• v.10 no.8
• /
• pp.319-330
• /
• 2021

Recently, in the field of open domain natural language question answering, multi-task, multi-hop question answering has been studied extensively. In this paper, we propose a novel deep neural network model using hierarchical graphs to answer effectively such multi-task, multi-hop questions. The proposed model extracts different levels of contextual information from multiple paragraphs using hierarchical graphs and graph neural networks, and then utilize them to predict answer type, supporting sentences and answer spans simultaneously. Conducting experiments with the HotpotQA benchmark dataset, we show high performance and positive effects of the proposed model.