감성분석을 활용한 수산물 소비 영향 분석
- Abstract
- This study explores how negative information about seafood impacts seafood consumption in South Korea by extrapolating the lagged effect that media coverage on Fukushima Daiichi nuclear power plant accident has on the domestic seafood consumption. The paper investigated mackerel based on a time series data set from January 2013 to December 2021, encompassing a total of 108 monthly observations. This study collected news articles from Naver News by utilizing the search terms “Fukushima” and “contaminated water”. Subsequently, a monthly sentiment index was derived from those articles by implementing a Long Short-Term Memory(LSTM) sentiment analysis model, which allowed for the classification of the articles into positive, negative, neutral polarities. In the empirical analysis, Polynomial Distributed Lag(PDL) Model was employed to estimate the lagged effects of sentiment index, domestic mackerel production, Norwegian mackerel imports, and domestic fresh and chilled mackerel prices on domestic mackerel consumption. Furthermore, impulse response analysis was applied to investigate the relationship between Norwegian import volume and sentiment index variable. The PDL analysis determined the media coverage, up to four months prior to the present moment, influences current consumption approximately by –1.2%. Accordingly, these findings demonstrated negative media coverage on seafood safety, such as the risk of radioactive contamination in seafood, has a long-term negative impact on mackerel consumption. Additionally, the impulse response analysis observed the shocks to the sentiment index influenced Norwegian mackerel imports, ranging from approximately +0.7 to +30.8%, two to six months after the shocks. Consequently, negative information about seafood has led to drop in domestic mackerel consumption, while in contrast the demand for Norwegian mackerel has risen.
- Author(s)
- 최연지
- Issued Date
- 2023
- Awarded Date
- 2023-08
- Type
- Dissertation
- Keyword
- Sentiment analysis, Mackerel, Seafood consumption, LSTM
- Publisher
- 부경대학교
- URI
- https://repository.pknu.ac.kr:8443/handle/2021.oak/33321
http://pknu.dcollection.net/common/orgView/200000693024
- Affiliation
- 부경대학교 대학원
- Department
- 대학원 해양수산경영학과
- Advisor
- 김도훈
- Table Of Contents
- I. 서론 1
1. 연구의 배경 및 목적 1
2. 연구의 범위 및 방법 3
II. 선행연구 5
1. 수산물 소비 관련 연구 5
2. 감성분석 관련 연구 11
III. 현황 분석 15
1. 후쿠시마 원전사고 15
2. 국내 수산물 수입구조 18
3. 주요 일본산 수입 어종 20
4. 국내 고등어 수급구조 23
IV. 이론적 배경 26
1. LSTM(Long Short-Term Memory) 딥러닝 모델 26
2. 다항시차분포모형(Polynomial Distributed Lag Model, PDL) 30
3. 벡터자기회귀모형(Vector Auto Regressive model, VAR) 35
V. 연구 설계 40
1. 데이터 수집 40
2. LSTM 모델 적용 45
Ⅵ. 분석 결과 53
1. 다항시차분포모형 분석 결과 53
2. 벡터자기회귀모형 및 충격반응함수 분석 결과 66
Ⅶ. 결론 및 제언 73
참고 문헌 78
- Degree
- Master
-
Appears in Collections:
- 대학원 > 해양수산경영학과
- Authorize & License
-
- Authorize공개
- Embargo2023-08-07
- Files in This Item:
-
Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.