簡易檢索 / 詳目顯示

回結果列表
研究生： 黎彥彣
Yan-Wun Lee
論文名稱： 基於多方面的情感分析模型之共同處理方法
A Co-processing Method for Aspect-based Sentiment Analysis Model
指導教授： 楊英魁
Ying-Kuei Yang
口試委員: 陳俊良
Jiann-Liang Chen
張博綸
Po-Lun Chang
李建南
Jian-nan Li
學位類別： 碩士
Master
系所名稱： 電資學院 - 電機工程系
Department of Electrical Engineering
論文出版年： 2019
畢業學年度： 107
語文別： 中文
論文頁數： 84
中文關鍵詞： 情感分析 、自然語言處理 、方面等級 、多方面共同處理 、機器學習
外文關鍵詞： aspect-level, multi-aspect co-processing
相關次數： 點閱：259  下載：20 
分享至:
查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報
  •  上一筆

    • 情感分析(sentiment analysis)又稱為意見探勘(opinion exploration)，是一門研究人
    類對於某物品或某事件的看法為正向還是負向的學問，它屬於自然語言處理領域中的
    一個副領域。這個領域可以應用於分析產品的好感度或在社交網站上分析民眾對於公
    共政策的想法，情感分析發展的目標就是以機器取代人力來完成判斷情感的動作。
    早期的情感分析方法以字典法為主，因為當時的硬體水準並沒有辦法執行那麼複
    雜的運算，只要將已知的情感詞語標記後存入字典中就可以用查字典的方式來實作情
    感分析。今日大多使用機器學習法來實作情感分析，因為機器學習法可以訓練機器舉
    一反三，也較容易跳脫出人類的思想框架來得到更好的預測結果。目前較常被應用於
    情感分析的模型是長短期記憶模型(long short-term memory, LSTM)，它擁有簡單的記
    憶功能，因此許多相關研究都是以它為基礎架構來建構自己的模型，本論文提出的模
    型也是基於長短期記憶模型的概念而延伸出來的。
    本論文提出一個新型的基於方面的情感分析模型，主要是設計來處理同一語句中
    擁有多個以上欲分析方面的情況，所以將它命名為多方面共同處理模型(multi-aspect
    co-processing model, MCM)。透過語義相容向量將語句整體的語義提取出來，並搭配
    因應不同的處理方面而有所不同的語義獨立向量來提取方面對於語句的語義，再結合
    專屬於方面的方面嵌入向量來表示情感代表向量。本論文提出的方法可以在多方面的
    情境時擁有較好的表現，雖然不能適用於各種情況，但終究還是個不錯的創新架構。

    Sentiment analysis is also being called as opinion exploration. It is a study of whether
    human beings are feeling positive or negative about an object or an event. It is also a
    subfield of natural language processing (NLP). This field actually works on some places
    like analyzing a product’s pros and cons or analyzing what public thought about the policy.
    The final purpose of sentiment analysis is to judge sentiments only by machine.
    The most popular methods proposed in the early stage of sentiment analysis are
    lexicon-based method because of the low performance of hardware at that time. We can
    easily run a lexicon-based method by comparing sentiment words which was already
    placed in dictionary and tagged by human. Today, most methods of sentiment analysis are
    machine-learning type methods because machines can infer beyond the limits of human
    beings. Nowadays, the most popular model being applied on sentiment analysis is long
    short-term model (LSTM). Because of the characteristic of being able to memorize
    previous experience, many related researches apply LSTM to be their fundamental
    structures. The model proposed in this thesis is also based on LSTM structure.
    A new aspect-based sentiment analysis method, called as multi-aspect co-processing
    model (MCM), is proposed in this thesis to solve the problems caused by co-processing
    multi-aspects. By taking out the meanings of a sentence, a sentiment representation vector
    can be formed by concatenating the meanings of the sentence and the meanings of its
    related aspect. Although the proposed model cannot work well in every occasion, it is still
    a valuable model in terms of solving the multi-aspect problem.

    摘要 ...... iv Abstract ...... v 第一章 緒論 .............. 1 1.1 研究背景與動機 ......... 1 1.2 研究目的 ........ 3 1.3 章節概要 ......... 3 第二章 文獻探討 ........ 4 2.1 情感分析 ........ 4 2.1.1 字典法(Lexicon-based) ........ 4 2.1.2 機器學習法(Machine Learning-based) ....... 5 2.2 詞向量(Word-embedding) ...... 5 2.2.1 Word2vec ......... 6 2.2.2 Global Vectors for Word Representation (Glove)....... 7 2.3 情感分析中常見的激發函數 .......... 8 2.3.1 S 形函數(Sigmoid function) ...... 8 2.3.2 雙曲正切函數(Hyperbolic tangent function, tanh) ........ 8 2.4 基於方面的情感分析所遭遇到的問題 ....... 9 2.4.1 方面嵌入向量平均問題(Aspect-vector average problem) ....... 9 2.4.2 單一代表向量問題(Representation vector simplification problem) ...... 9 2.4.3 方面類型問題(Aspect type problem) ........... 9 2.4.4 多方面共同處理問題(Multi-aspect co-processing problem) ...... 10 2.5 常用於情感分析的基礎模型 ......... 10 2.5.1 遞迴神經網路(Recurrent Neural Network, RNN) .......... 10 2.5.2 長短期記憶模型(Long Short-term Memory, LSTM) ........ 11 2.5.3 門控式遞迴單元(Gated Recurrent Unit, GRU) ........ 13 2.6 基於方面之情感分析的相關研究 ......... 16 2.6.1 Target-Dependent LSTM (TD-LSTM) ........ 16 2.6.2 Target-Connection LSTM (TC-LSTM) ....... 16 2.6.3 Aspect Embedding LSTM (AE-LSTM) ....... 17 2.6.4 Attention-based LSTM (AT-LSTM) ....... 17 2.6.5 Attention-based LSTM with Aspect Embedding (ATAE-LSTM) ........ 18 2.6.6 Deep Memory Network (MemNet) ....... 19 2.6.7 Feature-based Compositing Memory Networks (FCMN) ....... 20 2.7 研究目標 ........ 21 第三章 多方面共同處理之情感分析模型 .............. 22 3.1 設計理念 ........... 22 3.1.1 輸入資料的型式 ........... 22 3.1.2 多方面同時處理 ........ 23 3.1.3 方面嵌入向量的平均問題 ....... 27 3.2 輸入資料 ......... 30 3.2.1 語言的選擇 ........... 30 3.2.2 方面的提取方法 ......... 33 3.2.3 Semeval2014 資料集 ......... 35 3.2.4 轉換輸入資料的儲存方式 ........ 36 3.2.5 記錄方面詞語在語句中的位置 .......... 41 3.2.6 詞向量轉換 ....... 44 3.2.7 極性轉換 ........ 46 3.3 模型架構 .......... 47 3.3.1 LSTM 及平均單元 ........ 48 3.3.2 GRU ........ 50 3.3.3 串接單元 ....... 51 3.3.4 轉換矩陣 ........ 53 3.3.5 Softmax 分類器 ....... 53 3.4 訓練內容 .......... 55 3.4.1 損失函數 ........ 55 3.4.2 優化器 ....... 57 第四章 實驗結果分析與討論 ......... 59 4.1 開發環境 ........... 59 4.2 數值設定 ....... 59 4.2.1 參數設定 ....... 59 4.2.2 訓練變量 .......... 60 4.3 MCM 的實驗結果 ........ 61 4.3.1 與其他方法的準確度比較 ........ 61 4.3.2 多方面情境下的表現 ......... 62 4.3.3 極性對於正確率的影響 ...... 64 4.3.4 訓練時間的比較 ......... 65 第五章 結論與未來展望 ......... 67 5.1 結論 ........ 67 5.2 未來展望 ....... 69 參考文獻 ....... 70

    [1] W. J. Frawley, G. Piatesky-Shapiro, and C. J. Matheus, “Knowledge discovery in
    databases: an overview,” AI Magazine, vol. 13, no. 3, pp. 57-70, Sep. 1992.
    [2] A. Kao, and S. R. Poteet, Natural Language Processing and Text Mining, London:
    Springer, 2007.
    [3] N. V. Devi, and R. Ponnusamy, ”A systematic survey of natural language processing
    (NLP) approaches in different systems”, International Journal of Computer Sciences and
    Engineering, vol. 4, Issue. 7, pp. 192-198, Jul, 2016.
    [4] H. Singh, and A. K. Bathla, ”A survey on speech recognition”, International Journal of
    Advanced Research in Computer Engineering & Technology (IJARCET), vol. 2, Issue. 6,
    pp. 2186-2189, Jun, 2013.
    [5] C. D. Manning, and H. Schutze, Foundations of Statistical Natural Language Processing,
    1st ed. Cambridge, Massachusetts, London, England: M.I.T. Press, 1999.
    [6] F. Y. Y. Choi, “Advances in domain independent linear text segmentation,” in 1st Proc.
    North American Chapter of the Association for Computational Linguistics (NAACL),
    Manchester England, 2000, pp. 26-33.
    [7] P. J. Stone, D. C. Dunphy, and M. S. Smith, The general inquirer: A computer approach
    to content analysis, Oxford, England: M.I.T. Press, 1996.
    [8] M. Yang, Q. Qu, X. Chen, C. Guo, Y. Shen, and K. Lei, “Feature-enhanced attention
    network for target-dependent sentiment classification”, Neurocomputing, vol. 307, pp.
    91-97, 2018.
    [9] B. Liu, Sentiment Analysis and Opinion Mining, Morgan & Claypool Publishers,
    2012.
    [10] L. Jiang, M. Yu, M. Zhou, X. Liu, and T. Zhao, “Target-dependent Twitter sentiment
    classification,” in 49th Proc. Annual Meeting of the Association for Computational
    Linguistics (AMACL), Portland Oregon, Jun. 2011, pp. 151-160.
    [11] D. Tang, B. Qin, X. Feng, and T. Liu, “Effective LSTMs for target-dependent sentiment
    classification,” in 26th Proc. Conference on Computational Linguistics (COLING),
    Osaka Japan, Dec. 2016, pp. 3298-3307.
    [12] D. Tang, B. Qin, and T. Liu, “Aspect level sentiment classification with deep memory
    network,” in Proc. Empirical Methods in Natural Language Processing (EMNLP),
    Austin Texas, Nov. 2016, pp. 214-224.
    [13] Y. Wang, M. Huang, X. Zhu, and L. Zhao, “Attention-based LSTM for aspect-level
    sentiment classification,” in Proc. Empirical Methods in Natural Language Processing
    (EMNLP), Austin Texas, Nov. 2016, pp. 606-615.
    [14] L. Zhang, S. Wang, B. Liu, “Deep learning for sentiment analysis: A survey,” Data
    Mining and Knowledge Discovery, Mar. 2018.
    [15] A. Esuli, and F. Sebastiani, “Sentiwordnet: A publicly available lexical resource for
    opinion mining,” in 5th Proc. Language Resources and Evaluation (LREC), Genoa Italy,
    May. 2006, pp. 417-422.
    [16] I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning, M.I.T. Press, 2016.
    [17] Z. S. Harris, “Distributional structure”, WORD, vol. 10, issue. 2-3, pp. 146-162, Dec.
    1954.
    [18] S. Deerwester, S. T. Dumais, G. W. Furnas, T. K. Landauer, and R. Harshman,
    “Indexing by latent semantic analysis”, Journal of the American Society for
    Information Science (JASIS), vol. 41, Issue. 6, pp. 391-407, Sep. 1990.
    [19] T. Mikolov, K. Chen, G. Corrado, and J. Dean, “Efficient estimation of word
    representations in vector space,” in Proc. International Conference on Learning
    Representations (ICLR), Scottsdale Arizona USA, May. 2013, pp. 128-139.
    [20] T. Mikolov, I. Sutskever, K. Chen, G. Corrado, and J. Dean, “Distributed
    representations of words and phrases and their compositionality,” in 26th Proc.
    International Conference on Neural Information Processing Systems (NIPS), Lake
    Tahoe Nevada USA, vol. 2, Dec. 2013, pp. 3111-3119.
    [21] J. Pennington, R. Socher, and C. D. Manning, “Glove: Global vectors for word
    representation,” in Proc. Conference on Empirical Methods in Natural Language
    Processing (EMNLP), Doha Qatar, Oct. 2014, pp. 1532-1543.
    [22] G. Salton and M. McGill, Introduction to Modern Information Retrieval, New York,
    USA: McGraw-Hill, 1983.
    [23] J. L. Elman, “Finding structure in time”, Cognitive Science, vol. 14, Issue. 2, pp.
    179-211, Mar. 2016.
    [24] Z. Andreas, Simulation neuronaler Netze, Bonn: Addison-Weskey, 1994.
    [25] S. Hochreiter, J. Schmidhuber, “Long short-term memory”, Neural Computation, vol.
    9, Issue. 8, pp. 1735-1780, Nov. 1997.
    [26] K. Cho, B. Merrienboer, C. Gulcehre, D. Bahdanau, F. Bougares, H. Schwenk, and Y.
    Bengio, “Learning phrase representations using RNN encoder–decoder for statistical
    machine translation,” in Proc. Empirical Methods in Natural Language Processing
    (EMNLP), Doha Qatar, Oct. 2014, pp. 1724-1734.
    [27] R. Ma, K. Wang, T. Qiu, A. K. Sangaiah, D. Lin, and H. B. Liaqat, “Feature-based
    compositing memory networks for aspect-based sentiment classification in social
    internet of things”, Future Generation Computer Systems, vol. 92, pp. 879-888, 2019.
    [28] S. Bird, E. Klein, E. Loper, Natural Language Processing with Python: Analyzing Text
    with the Natural Language Toolkit, USA: O'Reilly Media, 2009.
    [29] S. Mohammad, S. Kiritchenko, and X. Zhu, “NRC-Canada: Building the
    state-of-the-art in sentiment analysis of tweets,” in 7th Proc. International Workshop
    on Semantic Evaluation (SemEval 2013), Atlanta Georgia USA, vol. 2, Jun. 2013, pp.
    321-327.
    [30] Z. Li, M. Zhang, S. Ma, B. Zhou, and Y. Sun, Information Retrieval Technology:
    Automatic extraction for product feature words from comments on the web, Berlin,
    Heidelberg: Springer, Oct. 2009, pp. 112-123.
    [31] H. Song, J. Chu, Y. Hu, and X. Liu, “Semantic analysis and implicit target extraction
    of comments from e-commerce websites,” in 4th Proc. World Congress on Software
    Engineering (WCSE), Hong Kong China, Dec. 2013, pp. 331-335.
    [32] K. W. Church, and P. Hanks, “Word association norms, mutual information, and
    lexicography”, Computational Linguistics, vol. 16, pp. 22-29, 1990.
    [33] M. Pontiki, D. Galanis, J. Pavlopoilos, H. Papagorgiou, I. Androutsopoulos, and S.
    Manandhar, “SemEval-2014 Task 4: aspect based sentiment analysis,” in 8th Proc.
    International Workshop on Semantic Evaluation (SemEval 2014), Dublin Ireland, Aug.
    2014, pp. 27-35.
    [34] K. P. Murphy, Machine Learning: A Probabilistic Perspective, M.I.T. Press, 2012.
    [35] D. P. Kingma, and J. L. Ba, “Adam: a method for stochastic optimization,” in 3rd Proc.
    International Conference on Learning Representations (ICLR), San Diego California,
    May. 2015.
    [36] D. E. Rumelhart, G. E. Hinton, and R. J. Williams, Neurocomputing: foundations of
    research, Cambridge, Massachusetts, USA: M.I.T. Press, 1988, pp. 696-699.
    [37] J. Duchi, E. Hazan, and Y. Singer, “Adaptive subgradient methods for online learning
    and stochastic optimization”, The Journal of Machine Learning Research, vol. 12, pp.
    2121-2159, 2011.
    [38] D. Ma, S. Li, X. Zhang, and H. Wang, “Interactive attention networks for aspect-level
    sentiment classification,” in 26th Proc. International Joint Conference on Artificial
    Intelligence (IJCAI), Melbourne Australia, Aug. 2017, pp. 4068-4074.
    [39] P. Chen, Z. Sun, L. Bing, and W. Yang, “Recurrent attention network on memory for
    aspect sentiment analysis,” in Proc. Empirical Methods in Natural Language
    Processing (EMNLP), Copenhagen Denmark, Sep. 2017, pp. 422-461.
    [40] Z. Zhang, L. Wang, Y. Zou, and C. Gan, “The optimally designed dynamic memory
    networks for targeted sentiment classification”, Neurocomputing, vol. 309, pp. 36-45,
    2018.
    [41] X. Ma, J. Zeng, L. Peng, G. Fortino, and Y. Zhang, “Modeling multi-aspects within
    one opinionated sentence simultaneously for aspect-level sentiment analysis”, Future
    Generation Computer Systems, vol. 93, pp. 304-311, 2019.

    QR CODE