透過您的圖書館登入
IP:3.133.133.110
  • 學位論文

基於深度學習的佛教引用句推薦系統

Building a Buddhist Quotes Recommendation System Based on Deep Learning Approaches

指導教授 : 洪振洲
共同指導教授 : 王昱鈞(yu-chun wang)

摘要


在撰寫文章時,我們時常會引經據典,引用一些名句以增強文章的說服力。所謂的佛教引用句指的是常見的佛教用語或經典名句,佛教引用句的哲學常給人心一種啟發作用,例如「凡所有相,皆是虛妄」、「應無所住,而生其心」「一切有為法,如夢幻泡影」等。因此於文章寫作時,如果能引用合適的佛教引用句,將可以讓文章更加生動有力。然而,對於一般人來說,要深入理解並熟知佛教引用句並不容易,此時如有一個推薦的方法,自動從文章的內容進行分析後,推薦合適的佛教引用句,寫作時更能得心應手。 在本文中,我們提出寫作時引用佛教引用句的推薦方法。我們嘗試解決這個吸引人的 AI任務,並建立名言推薦的學習框架;首先,從網路上收集比較通俗的佛教引用句,並挖掘包含這些佛教引用句的文章,以建立一個用於實驗的資料集。再者,實驗以監督學習方式運用了詞嵌入(Embedding)、深度學習(Deep Learning)方法為這項任務建立模型,用以分析理解佛言名句特徵與佛教引用句上下文特徵之間的相關性進行判斷,最後輸出佛教引用句推薦的引用。實驗結果顯示,我們提出的方法對於這項任務是合適的且準確度達91.48%。

並列摘要


In writing an article, we often quote maxims or some famous sayings to enhance the persuasiveness of the article. Buddhist quotation is a vital and meaningful source of the maxims. The so-called Buddhist quotation refers to the common Buddhist expressions or famous classic sayings. The philosophy of famous Buddhist quotations often makes people enlightened. Therefore, when writing an article, if we can quote the appropriate Buddhist maxim, it will make the article more vivid and powerful. However, for the average person, understanding and being familiar with famous Buddhist quotations is not easy, so a recommended method is necessary to automatically analyze the article's content and then recommend the appropriate Buddhist quotations. In this article, we propose and address the Buddhist quotation recommendation method to suggest famous Buddhist maxims. We adopt a deep learning approach to establish the quotation recommendation system. First, famous Buddhist maxims are selected, and the articles containing these Buddhist maxims are collected to construct a dataset for experimentation. Then, the supervised learning model comprised of word embedding and deep learning methods is established to generate a ranking list of appropriate Buddhist quotations based on the input context. The recommendation accuracy of our proposed method achieves 91.48%, that shows our method is effective for this quotation recommendation task.

參考文獻


參考資料
(一) 中文專書
1. Francois Chollet著、葉欣睿譯(2019),《Deep Learning深度學習必讀-Keras大神帶你用Python實做》,台北市:旗標科技。
2. 林大貴(2017),《TensorFlow+Keras深度學習人工智慧實務應用》,新北市:博碩文化。
3. 斎藤康毅著、吳嘉芳譯(2017),《Deep Learning:用Python進行深度學習的基礎理論實作》,台北市:碁峰。

延伸閱讀