基於少數關鍵字之半監督式學習法進行評論文件分類

Semi-supervised Review Classification with a Few Polarity Keywords

陳雅婷(Ya-Ting Chen)

指導教授 ： 鄭卜壬

國立臺灣大學/電機資訊學院/資訊工程學研究所/碩士(2013年)

https://doi.org/10.6342/NTU.2013.00250

摘要

評論文件分類(Review Classification)跟情緒分析有些類似，而情緒分析(Sentiment Analysis)主要是探討撰寫者的情緒狀態，具有高度領域相關(Domain Dependent)的特性。評論文件的極性分類最主要的問題是希望能夠自動化的分類那些沒有被標記過的評論文件極性，針對不同領域的文章進行分析，可能有不同的結果。本論文利用少量分類的關鍵字進行文件分類，使用兩種完全不同領域的語料進行研究，探討不同領域的代表詞彙，進而分類文件極性，實驗結果顯示分類效能有不錯的效果。

關鍵字

評論文件分類 ； 情緒分析 ； 關鍵字

並列摘要

Review classification and sentiment analysis are similar. Sentiment analysis mainly aims at exploring the emotional state of writers. The analysis highly depends on the application domains. The goal of review classification is the task of automatically classifying unlabeled documents. Analyzing polarity of the articles in different domains may have different results. In this study, we focus on two different domains of data, and use a few positive and negative keywords about that domain to classify the sentiment of articles. The experiments show that the proposed methods have a better classification performance.

並列關鍵字

Review Classification ； Sentiment Analysis ； Category Keyword

參考文獻

[1] HC Yu, TH Huang, HH Chen, “Domain Dependent Word Polarity Analysis for Sentiment Classification”, ROCLING, 2012.

[5] CC Chang and CJ Lin. “Libsvm: A Library for Support Vector Machines.” ACM Transactions on Intelligant Systems and Technology, 2:27:1-27:27,2011.

[10] Chaovalit P. and Zhou, L., “Movie Review Mining: a Comparison between Supervised and Unsupervised”, Proceedings of the 38th Hawaii International Conference on System Sciences, pp.112c-112c, 2005.

[11] Zhongchao Fei; Jian Liu; Gengfeng Wu; “Sentiment Classification Using Phrase Patterns ”, Computer and Information Technology, 2004. CIT’04. The Fourth International Conference on , vol., no., pp.1147 –1152, 14-16 Sept. 2004.

[12] Joachims, T., “Text Categorization with Support Vector Machines: Learning with Many Relenvant Features”, Proceedings of th Eourpean Conference on Machine Learning, pp. 21-24 (pp. 137-142), 1998.