跳到主要內容

臺灣博碩士論文加值系統

(100.28.132.102) 您好!臺灣時間:2024/06/23 07:52
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果 :::

詳目顯示

我願授權國圖
: 
twitterline
研究生:陳彥亨
研究生(外文):Yan-Heng Chen
論文名稱:利用社群網站使用者互動關係與情感意向進行民意意向分析
論文名稱(外文):Public Opinion Tendency Analysis by Combining User Relationship and Emotional Opinion in Social Networks
指導教授:王正豪王正豪引用關係
口試委員:劉傳銘王正豪楊凱翔
口試日期:2018-07-19
學位類別:碩士
校院名稱:國立臺北科技大學
系所名稱:資訊工程系
學門:工程學門
學類:電資工程學類
論文種類:學術論文
論文出版年:2018
畢業學年度:107
語文別:中文
論文頁數:37
中文關鍵詞:卷積神經元網路(CNN)(Convolutional Neural Networks)支援向量機(SVM)(Support Vector Machine)情感意向民意預測
外文關鍵詞:Convolutional Neural NetworksSupport Vector MachineEmotional opinionOpinion prediction
相關次數:
  • 被引用被引用:0
  • 點閱點閱:160
  • 評分評分:
  • 下載下載:0
  • 收藏至我的研究室書目清單書目收藏:0
社群網站成為現今人們抒發自我情緒意見的平台,參與者大多是關心時事的民眾,或一些在生活中較無和群體互動的隱性群眾。民意意向分析的常見作法有:情感分析,以及使用者關係兩方面。在情感意向分析上,傳統方法大多只分析字詞,常常未考慮到句子間的關係,例如:"我很喜歡這件衣服,但是我的預算不足"。上述這兩句話在傳統方法會根據裡面的字詞"喜歡"和"預算不足"進行分析,而未考慮到這兩句話中間有個"但是"字詞會讓兩句話有關聯。在使用者關係方面,大部分以發文和回覆的使用者互動關係來探討,但是會有一些隱性的關係未被考慮到,例如:回覆者回文時不只和發文者建立關係,並且會在內容中提及其他使用者。
本論文結合使用者發文內容的情感分析及參與討論者互動關係,進行民意意向分析。情感分析的部分以Doc2Vec(Document to vector)的word embedding來進行文章特徵擷取,結合GlobalTextCNN分類器,來訓練文章的情感意向,此方法不僅考慮字與字之間的關係,也探討句子與字之間的關係。再利用發文者和參與者的互動關係,如社群平台文章的喜歡(like)、轉推(retweet)和提及(mention)等為特徵,來探討人與人之間的關係,這樣可以有較多元性的連結,做為發文者的權重,進行民意意向分析。
根據Twitter的實驗結果顯示,結合Doc2Vec和GlobalTextCNN的F1-measure分數達到88.85%,並且利用2017年英國大選來進行民意意向進行實驗,而結果顯示MAE平均絕對誤差到達5.0%,驗證了本方法的有效性。
The social networking site has become a platform for people to express their own emotional opinions. Participants are mostly people who care about current events, or those who do not interact with groups in life. Common ways in the analysis of public opinion tendency include emotional analysis and user relations. In the analysis of emotional opinion, traditional word embedding methods are relationship between words, so the relationship between sentences is often not taken into consideration. For example, "I like this dress very much, but my budget is insufficient". The above two sentences are analyzed in the traditional way according to the words "like" and "insufficient budget", but without considering that there is a "but" word between the two sentences that will make the two sentences related. Most of user relationships are discussed between users who post articles and user who respond with others, but there are some hidden relationships that are not considered. For example, when a user replies, it not only establishes a relationship with the author, but also mentions other users in the content. In this paper, we combine the emotional analysis of the users published content with the interaction of the participants to conduct the analysis of public opinion. The part of emotional analysis uses Doc2Vecs word embedding to extract the features of the article, and combines the GlobalTextCNN classifier to train the sentiment orientation of the emotional opinion of the article. This method not only considers the relationship between words and words, but also discusses the relationship between sentences and words. We use the interaction between the author and the participants. For example, the like, retweet, and mention of social media platform articles are used to explore the relationship between people. By considering more possible ways of connection, and the weights of users can be estimated when analyzing the opinion of the public.
According to the experimental results of emotional opinion analysis, F1-measure score which combined with Doc2Vec and GlobalTextCNN is 88.85%. And using the 2017 British election to conduct an public opinion tendency’s experiment, the results showed that MAE is 5.0%. This verifies the effectiveness of the proposed approach in public opinion analysis.
摘要 i
Abstract iii
誌謝 v
目錄 vi
圖目錄 viii
表目錄 ix
第一章 緒論 1
1.1 研究背景與動機 1
1.2 研究目的 1
1.3 研究貢獻 2
1.4 章節概要 3
第二章 相關研究 4
2.1 社群網站民意與分析 4
2.2 使用者互動關係分析 5
2.2.1 PageRank 5
2.2.2 使用者互動關係種類 6
2.3 情感意向分析 6
2.3.1 情感意向方法 7
2.3.2 以類神經網路進行情感分析 8
第三章 研究方法 10
3.1 方法概述 10
3.2 社群網站文章 11
3.3 使用者互動關係分析 11
3.3.1 使用者互動關係圖建構 11
3.3.1.1 互動關係連結 12
3.3.1.2 互動關係圖節點 13
3.3.2 使用者互動分數計算 14
3.4 情感意向分析 16
3.4.1 Word Embedding 16
3.4.2 情感意向分類 19
3.4.2.1 分類模型訓練 19
3.4.2.2 情感意向預測 22
3.4.3 使用者民意意向計算 22
3.5 民意意向分數彙整 23
第四章 實驗與討論 25
4.1 實驗環境 25
4.2 實驗資料蒐集 25
4.3 使用者互動關係分析實驗 25
4.4 情感意向分析實驗 27
4.5 民意意向分析結果 31
第五章 結論 34
參考文獻 35
[1] Yamaguchi, Yuto, Tsubasa Takahashi, Toshiyuki Amagasa, and Hiroyuki Kitagawa. Turank: Twitter User Ranking Based on User-Tweet Graph Analysis. International Conference on Web Information Systems Engineering: Springer, 2010.
[2] Diakopoulos, Nicholas A and David A Shamma. Characterizing Debate Performance Via Aggregated Twitter Sentiment. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems: ACM, 2010.
[3] Bermingham, Adam and Alan Smeaton. On Using Twitter to Monitor Political Sentiment and Predict Election Results. Proceedings of the Workshop on Sentiment Analysis where AI meets Psychology (SAAIP 2011), 2011.
[4] Burnap, Pete, Rachel Gibson, Luke Sloan, Rosalynd Southern, and Matthew Williams. "140 Characters to Victory?: Using Twitter to Predict the Uk 2015 General Election." Electoral Studies 41, 2016.
[5] Min, Qiang , and Shuicheng . "Network in Network." arXiv preprint arXiv:1312.4400, 2013.
[6] Grčar, Miha, Darko Cherepnalkoski, Igor Mozetič, and Petra Kralj Novak. "Stance and Influence of Twitter Users Regarding the Brexit Referendum." Computational social networks 4, no. 1, 2017.
[7] Hirsch, Jorge E. "An Index to Quantify an Individuals Scientific Research Output." Proceedings of the National academy of Sciences 102, no. 46, 2005.
[8] Mao, Guo-Jun and Jie Zhang. A Pagerank-Based Mining Algorithm for User Influences on Micro-Blogs. PACIS, 2016.
[9] Page, Lawrence, Sergey Brin, Rajeev Motwani, and Terry Winograd. The Pagerank Citation Ranking: Bringing Order to the Web. Stanford InfoLab, 1999.

[10] Alhelbawy, Ayman and Robert Gaizauskas. Graph Ranking for Collective Named Entity Disambiguation. Vol. 2. Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), 2014.
[11] Cha, Meeyoung, Hamed Haddadi, Fabricio Benevenuto, and P Krishna Gummadi. "Measuring User Influence in Twitter: The Million Follower Fallacy." Icwsm 10, no. 10-17 , 2010.
[12] Kato, Shoko, Akihiro Koide, Takayasu Fushimi, Kazumi Saito, and Hiroshi Motoda. Network Analysis of Three Twitter Functions: Favorite, Follow and Mention. Pacific Rim Knowledge Acquisition Workshop: Springer, 2012.
[13] Singh, Vivek Kumar, Rajesh Piryani, Ashraf Uddin, and Pranav Waila. Sentiment Analysis of Movie Reviews: A New Feature-Based Heuristic for Aspect-Level Sentiment Classification. Automation, computing, communication, control and compressed sensing (iMac4s), 2013 international multi-conference on: IEEE, 2013.
[14] Thelwall, Mike. "The Heart and Soul of the Web? Sentiment Strength Detection in the Social Web with Sentistrength." In Cyberemotions, 119-34: Springer, 2017.
[15] Kouloumpis, Efthymios, Theresa Wilson, and Johanna D Moore. "Twitter Sentiment Analysis: The Good the Bad and the Omg!" Icwsm 11, no. 538-541, 2011.
[16] Kang, Hanhoon, Seong Joon Yoo, and Dongil Han. "Senti-Lexicon and Improved Naïve Bayes Algorithms for Sentiment Analysis of Restaurant Reviews." Expert Systems with Applications 39, no. 5, 2012.
[17] Yin, Wenpeng, Katharina Kann, Mo Yu, and Hinrich Schütze. "Comparative Study of Cnn and Rnn for Natural Language Processing." arXiv preprint arXiv:1702.01923, 2017.
[18] Min, Qiang, and Shuicheng Yan. "Network in Network." arXiv preprint arXiv:1312.4400, 2013.
[19] Kim, Yoon. "Convolutional Neural Networks for Sentence Classification." arXiv preprint arXiv:1408.5882 , 2014.
[20] Mikolov, Tomas, Kai , Greg Corrado, and Jeffrey Dean. "Efficient Estimation of Word Representations in Vector Space." arXiv preprint arXiv:1301.3781, 2013.
[21] Quoc and Tomas Mikolov. Distributed Representations of Sentences and Documents. International Conference on Machine Learning, 2014.
[22] Zhang and Byron Wallace. "A Sensitivity Analysis of (and Practitioners Guide to) Convolutional Neural Networks for Sentence Classification." arXiv preprint arXiv:1510.03820 , 2015.
[23] Chih-Chung and Chih-Jen "Libsvm: A Library for Support Vector Machines." ACM transactions on intelligent systems and technology (TIST) 2, no. 3, 2011.
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top