臺灣博碩士論文加值系統

本研究旨在建立以深度學習應用於財經語句情緒極性之分類模型，並且結合中文財經領域情緒辭典以強化模型對於情緒詞之特徵感知能力，以台灣財經新聞為資料集，將財經新聞中語句之情緒極性作為本研究之分析對象，所建立之模型將用來分析財經新聞中各語句之情緒極性，並且與國外以深度學習方式結合情緒辭典之相關研究所建立之模型進行比較，以證實本研究所使用之方法對於台灣財經新聞語句情緒極性分類之貢獻。

This study intends to establish a deep learning model of sentiment polarity analysis applied in financial texts, and combines the emotional dictionary of finance to strengthen the model's ability to perceive the characteristics of emotional words, and use Taiwan financial news as a data set to express the emotional polarity of statements in financial news. As the object of analysis of this study, the established deep learning model will be used to analyze the emotional polarity of each sentence in the financial news, and compare it with the model established by the relevant research institutes of deep learning with using emotional dictionary in foreign countries to confirm the study contributions in this field.

目次
謝誌 2
摘要 3
Abstract 4
目次 i
圖次 iii
表次 iv
第一章 緒論 1
第一節 研究背景與研究動機 2
第二節 研究內容與範圍 3
第三節 論文架構 4
第二章 文獻探討 5
第一節 文字探勘 5
第二節 文字情緒極性分析 7
第三節 文字特徵表示 8
第四節 深度學習（Deep Learning） 14
第五節 財經情緒辭典 26
第三章 研究方法 31
第一節 中文財經新聞資料蒐集 31
第二節 資料前處理 32
第三節 詞向量準備 38
第四節 情緒極性標記 39
第五節 建構模型 46
第六節 模型評估 53
第四章 研究結果與討論 55
第一節 實驗與驗證方法 55
第二節 實驗結果分析 58
第三節 模型注意力機制視覺化 62
第四節 模型應用於其他資料集 64
第五章 結論與建議 67
第一節 研究結論 67
第二節 研究限制 67
第三節 研究建議 68
參考文獻 71

[1] Liddy, E.D., Natural language processing. 2001.
[2] Navigli, R. Natural Language Understanding: Instructions for (Present and Future) Use. in IJCAI. 2018.
[3] 臺灣大學會計學研究所學位論文, 林.J., 財經領域情緒辭典之建置與其有效性之驗證-以財經新聞為元件. 2013: p. 1-60.
[4] 張津挺, 中文財務情緒字典建構與其在財務新聞分析之應用 = On the Construction and Analysis of Chinese Financial Sentiment Lexicon for Financial News. 2016: 臺北市立大學 ;.
[5] Shin, B., T. Lee, and J.D.J.a.p.a. Choi, Lexicon integrated cnn models with attention for sentiment analysis. 2016.
[6] Frawley, W.J., G. Piatetsky-Shapiro, and C.J.J.A.m. Matheus, Knowledge discovery in databases: An overview. 1992. 13(3): p. 57.
[7] Fayyad, U., G. Piatetsky-Shapiro, and P.J.C.o.t.A. Smyth, The KDD process for extracting useful knowledge from volumes of data. 1996. 39(11): p. 27-34.
[8] Harris, Z.S.J.W., Distributional structure. 1954. 10(2-3): p. 146-162.
[9] Firth, J.R.J.S.i.l.a., A synopsis of linguistic theory, 1930-1955. 1957.
[10] Mikolov, T., et al., Efficient estimation of word representations in vector space. 2013.
[11] LeCun, Y., Y. Bengio, and G.J.n. Hinton, Deep learning. 2015. 521(7553): p. 436.
[12] Hochreiter, S. and J.J.N.c. Schmidhuber, Long short-term memory. 1997. 9(8): p. 1735-1780.
[13] Olah, C.J.G.b., posted on August, Understanding lstm networks. 2015. 27: p. 2015.
[14] Goo, C.-W.J.臺., 雙層語意控制之自然語言理解與自動對話摘要. 2018: p. 1-45.
[15] Vaswani, A., et al. Attention is all you need. in Advances in neural information processing systems. 2017.
[16] Devlin, J., et al., BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. 2018.
[17] Leskovec, J., A. Rajaraman, and J.D. Ullman, Mining of massive datasets. 2014: Cambridge university press.
[18] Mikolov, T., et al. Distributed representations of words and phrases and their compositionality. in Advances in neural information processing systems. 2013.
[19] Koppel, M. and J.J.C.I. Schler, The importance of neutral examples for learning sentiment. 2006. 22(2): p. 100-109.
[20] Fleiss, J.L.J.P.b., Measuring nominal scale agreement among many raters. 1971. 76(5): p. 378.
[21] Landis, J.R. and G.G.J.b. Koch, The measurement of observer agreement for categorical data. 1977: p. 159-174.
[22] Kim, Y.J.a.p.a., Convolutional neural networks for sentence classification. 2014.
[23] Chang, C.-C., C.-J.J.A.t.o.i.s. Lin, and technology, LIBSVM: a library for support vector machines. 2011. 2(3): p. 27.
[24] Pang, B. and L. Lee. Seeing stars: Exploiting class relationships for sentiment categorization with respect to rating scales. in Proceedings of the 43rd annual meeting on association for computational linguistics. 2005. Association for Computational Linguistics.
[25] Baccianella, S., A. Esuli, and F. Sebastiani. Sentiwordnet 3.0: an enhanced lexical resource for sentiment analysis and opinion mining. in Lrec. 2010.