BERT集成學習方法於財經文本探勘之論點驅動的情感分析任務

English | 正體中文 | 简体中文 | Items with full text/Total items : 45422/58598 (78%)
Visitors : 2515161 Online Users : 233

RC Version 7.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.

Home ‧ Login ‧ Upload ‧ Help ‧ About ‧ Administer

TMUIR > College of Management > Graduate Institute of Data Science > Dissertations/Theses > Item 987654321/64313

Please use this identifier to cite or link to this item: http://libir.tmu.edu.tw/handle/987654321/64313

Title:	BERT集成學習方法於財經文本探勘之論點驅動的情感分析任務 Ensemble BERT Approach for Argument-Driven Sentiment Analysis in Financial Text Mining Applications
Authors:	施辰穎 SY, EUGENE LALIS
Contributors:	大數據科技及管理研究所碩士班張詠淳
Keywords:	Financial NLP、Argumentative Mining、Ensemble Technique、Voting Mechanism、Argument Unit Classification
Date:	2024-06-17
Issue Date:	2024-09-30 14:21:00 (UTC+8)
Abstract:	Financial discourse contains complex arguments and sentiments that significantly impact market trends and strategic decisions. This thesis presents a novel ensemble approach to argument mining, advancing the state-of-the-art in argument unit identification for the financial domain. By integrating multiple fine-tuned pre-trained language models through robust voting, the pro-posed ensemble architecture effectively leverages diverse model strengths while mitigating in-dividual biases. Rigorous experimentation on the FinArg dataset demonstrated the ensemble's superiority, with the top configuration achieving a 77.083% Macro F1-score, outperforming the best individual model by 0.659%. Comprehensive analysis revealed insights into how voting techniques impact generalization and the trade-offs between flexibility and robustness. Moreo-ver, this work pioneers integrating argument mining with financial sentiment analysis. Incorpo-rating argument probabilities as features into FinancialBERT enhanced sentiment prediction on the Financial Phrase Bank dataset, with Macro F1-scores reaching 87.818% (FPBall) and 96.919% (FPB100). Strong argument-sentiment correlations open avenues for leveraging argu-mentative structure in sentiment analysis. Contributions include the innovative ensemble archi-tecture, insights into ensemble voting dynamics, and pioneering argument-sentiment integration, enabling sophisticated analytical tools for applications like risk assessment and market fore-casting.
Description:	碩士指導教授：張詠淳口試委員：張詠淳口試委員：陳建錦口試委員：蘇家玉
Note:	論文公開日期：2024-07-11
Data Type:	thesis
Appears in Collections:	[Graduate Institute of Data Science] Dissertations/Theses

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	66	View/Open

著作權聲明 Copyright Notice

本平台之數位內容為臺北醫學大學所收錄之機構典藏，包含體系內各式學術著作及學術產出。秉持開放取用的精神，提供使用者進行資料檢索、下載與取用，惟仍請適度、合理地於合法範圍內使用本平台之內容，以尊重著作權人之權益。商業上之利用，請先取得著作權人之授權。
The digital content on this platform is part of the Taipei Medical University Institutional Repository, featuring various academic works and outputs from the institution. It offers free access to academic research and public education for non-commercial use. Please use the content appropriately and within legal boundaries to respect copyright owners' rights. For commercial use, please obtain prior authorization from the copyright owner.
瀏覽或使用本平台，視同使用者已完全接受並瞭解聲明中所有規範、中華民國相關法規、一切國際網路規定及使用慣例，並不得為任何不法目的使用TMUIR。
By utilising the platform, users are deemed to have fully accepted and understood all the regulations set out in the statement, relevant laws of the Republic of China, all international internet regulations, and usage conventions. Furthermore, users must not use TMUIR for any illegal purposes.
本平台盡力防止侵害著作權人之權益。若發現本平台之數位內容有侵害著作權人權益情事者，煩請權利人通知本平台維護人員([email protected])，將立即採取移除該數位著作等補救措施。
TMUIR is made to protect the interests of copyright owners. If you believe that any material on the website infringes copyright, please contact our staff([email protected]). We will remove the work from the repository.

Loading...