자연어 처리를 이용한 배터리 소재 물성값 추출 자동화
- Alternative Title
- Automatic Extraction of Battery Materials properties Using Natural Language Processing
- Abstract
- Batteries play an essential role in various fields of modern society. The duration of device usage and the speed of battery charging depend on the battery's capacity and performance. Particularly, the anode material, a key component in storing lithium ions, critically affects the charging speed and lifespan of the battery. Consequently, battery-related research is focused on developing new anode materials that offer higher capacities and improved performance. However, experimenting with different properties of batteries for the development of anode materials can be costly and pose safety risks. Additionally, the lack of systematic organization of numerous research findings often leads to redundant experiments. While referring to academic papers to obtain necessary property values is a solution, manual search and analysis by individuals have limitations and risk missing crucial information. To address these issues, this study has developed an automated model using Natural Language Processing (NLP) technology. It automatically extracts and organizes the names of materials and property values from a vast number of research papers on battery anode materials. This allows researchers to quickly identify the core topics of papers, clearly understand the content, track research trends, and swiftly grasp key keywords.
- Author(s)
- 김지연
- Issued Date
- 2024
- Awarded Date
- 2024-02
- Type
- Dissertation
- Keyword
- 자연어 처리 기술
- Publisher
- 국립부경대학교 대학원
- URI
- https://repository.pknu.ac.kr:8443/handle/2021.oak/33763
http://pknu.dcollection.net/common/orgView/200000743797
- Alternative Author(s)
- KIM JIYEON
- Affiliation
- 국립부경대학교 대학원
- Department
- 대학원 에너지자원공학과
- Advisor
- 여병철
- Table Of Contents
- 1. 서론 1
2. 연구 방법 4
2.1 데이터 수집 4
2.2 데이터 전처리 8
2.2.1 파싱 8
2.2.2 토큰화 10
2.3 데이터 시각화 11
2.3.1 Doc2Vec 11
2.3.2 t-SNE와 PCA 14
2.4 데이터 태깅 16
2.4.1 품사 태깅 16
2.4.2 B-I-O 태깅 18
2.5 자연어 처리 모델 20
2.5.1 BiLSTM 모델 20
2.5.2 BiLSTM-CRF 모델 22
2.5.3 BERT 모델 24
3. 연구 결과 26
3.1 데이터 태깅 결과 26
3.2 데이터 시각화 결과 29
3.3 자연어 처리 모델 학습 결과 33
4. 토의 35
5. 결론 36
참고 문헌 37
- Degree
- Master
-
Appears in Collections:
- 대학원 > 에너지자원공학과
- Authorize & License
-
- Authorize공개
- Embargo2024-02-16
- Files in This Item:
-
Items in Repository are protected by copyright, with all rights reserved, unless otherwise indicated.