메뉴 건너뛰기
.. 내서재 .. 알림
소속 기관/학교 인증
인증하면 논문, 학술자료 등을  무료로 열람할 수 있어요.
한국대학교, 누리자동차, 시립도서관 등 나의 기관을 확인해보세요
(국내 대학 90% 이상 구독 중)
로그인 회원가입 고객센터 ENG
주제분류

추천
검색
질문

논문 기본 정보

자료유형
학술대회자료
저자정보
박지연 (홍익대학교) 이재욱 (홍익대학교) 최제범 (홍익대학교) 홍서윤 (홍익대학교) 홍성준 (홍익대학교) 정은성 (홍익대학교)
저널정보
대한전자공학회 대한전자공학회 학술대회 2024년도 대한전자공학회 하계학술대회 논문집
발행연도
2024.6
수록면
2,580 - 2,584 (5page)

이용수

표지
📌
연구주제
📖
연구배경
🔬
연구방법
🏆
연구결과
AI에게 요청하기
추천
검색
질문

초록· 키워드

오류제보하기
The College Scholastic Ability Test (CSAT) has established itself as a crucial examination for assessing students" academic achievements. However, due to time constraints, maintaining the quality and diversity of English exam questions while providing a sufficient number of them can be challenging. This study aims to alleviate the constraints faced by question writers and provide effective learning experiences for students by developing a Large Language Model (LLM) to automatically generate CSAT English exam questions. To achieve this, data from CSAT English exams over the past decade and mock exam questions were utilized. The data was formatted according to directive adjustment formats and combined with predefined prompts for question generation. Due to limited resources, the model was trained using LoRA and quantization methods. The performance of the model was evaluated by comparing the generated questions with existing ones and conducting a survey with 10 participants based on predefined metrics. The evaluation results showed that for the items of ‘Inserting a sentence into the blank’ and ‘Author"s claim’, uniqueness criterion wasn"t met due to the p-value of Mann Whitney U test being lower than the significance level, resulting in the absence of a unique correct answer choice. Excluding this, for the items of ‘Inserting a sentence into the blank’, ‘Inference of the title’, ‘Author"s claim’, difficulty, fluency, and type consistency criteria, as well as for the uniqueness criterion of ‘Inference of the title’, the pvalue exceeded the significance level, indicating that the model generated questions with unique correct choices, appropriate difficulty, minimal grammatical errors, and answer choices consistent with the question type. This study is expected to contribute to saving costs and time for exam administrators and educational institutions, while providing learners with a more comprehensive learning experience through diverse types of questions.

목차

Abstract
Ⅰ. 서론
Ⅱ. 배경
Ⅲ. 본론 및 구현
Ⅳ. 실험 및 평가
Ⅴ. 결론 및 향후 연구 방향
참고문헌

참고문헌 (0)

참고문헌 신청

함께 읽어보면 좋을 논문

논문 유사도에 따라 DBpia 가 추천하는 논문입니다. 함께 보면 좋을 연관 논문을 확인해보세요!

이 논문과 함께 이용한 논문

최근 본 자료

전체보기

댓글(0)

0