現今，因為網路的快速發展，發展出了以使用者為中心的網站，並且提供網站(部落格、論壇、討論區…等)讓使用者分享自己對公司或產品的評價；在這些散布在網路上大量的文章，為了能快速取得文章內之資訊，中文意見探勘系統就為之重要。本論文主要研究中文意見探勘系統之句型文法，分析Mobile01和PTT的電信和網路兩大熱門討論區發文者對公司或產品之文章。我們句型文法主要利用「預設Topic」、「子句優先」和「對應關係」搭配句型來配對意見元素表達發文者意見。實驗結果顯示各月份和不同討論區之準確率、回收率和F1相差不多，顯示我們的句型文法是穩定的；另外在考慮預設Topic情況下，實驗結果顯示對整體之數據是有成效的。

Today, the fast development of the Web has resulted in the user-centric websites and websites (blog, forum, BBS, etc) that allow users to share their comments on a company or a product. For the rapid access to information contained in the huge amounts of articles on the Web, the Chinese opinion mining system is very important. This paper discusses the sentence grammar of the Chinese opinion mining system, analyzes the articles on companies or products published on two popular BBS of Mobile01 and PTT telecommunications. Our sentence grammar mainly uses “default topic”, “clause priority” and “corresponding relationship” coupled with sentences to match up with the opinion elements to express the views of the article posters. The experimental results suggest that precision, recall and F1 are the same in various months and BBS, suggesting our sentence grammar is stable. By considering the default topic, the experimental results prove that the proposed method is effective in terms of overall data.

目錄
第1章	緒論	1
1.1	研究動機與目的	1
1.2	研究架構	2
第2章	相關文獻	4
2.1	意見元素	4
2.2	Feature和Opinion Word的配對	5
2.3	對應關係	6
2.4	意見極性的判斷	7
第3章	句型文法	9
3.1	預設Topic	9
3.2	意見元素類別轉換	13
3.3	意見元素的配對方法	17
3.4	句型的配對	23
3.5	意見極性的判斷	35
第4章	實驗討論	38
4.1	資料集和評估方式	38
4.2	實驗結果	40
4.3	文章錯誤筆數分析	42
4.4	影響準確率之探討分析	46
4.5	影響回收率之探討分析	51
第5章	結論與未來展望	56
參考文獻		57
附錄-英文論文	59

圖目錄
Figure 1 短篇文章預設Topic之演算法	11
Figure 2 短篇文章預設Topic之範例	12
Figure 3 長篇文章預設Topic之範例	13
Figure 4 意見元素轉換Feature之範例	14
Figure 5 意見元素無轉換Feature之範例	15
Figure 6 意見元素轉為副詞之範例	15
Figure 7 意見元素無轉為副詞之範例	16
Figure 8 配對流程圖	24
Figure 9 一般句配對之範例	25
Figure 10 轉折句配對之範例	27
Figure 11 對等句配對之範例	29
Figure 12 比較句配對之範例一	30
Figure 13 比較句配對之範例二	31

表目錄
Table 1 意見元素	4
Table 2 詞庫表	19
Table 3 Feature和Opinion Word(OP)關係表	20
Table 4 Topic和Product關係表	20
Table 5 轉折詞表	26
Table 6 連接詞表	28
Table 7 疑問詞表	34
Table 8 假設詞表	34
Table 9 否定詞表	37
Table 10 結果聯立	38
Table 11 Mobile01電信各月整體數據	40
Table 12 Mobile01網路各月整體數據	40
Table 13 PTT電信各月整體數據	41
Table 14 PTT網路各月整體數據	41
Table 15 Mobile01電信各月影響準確率原因筆數	42
Table 16 Mobile01網路各月影響準確率原因筆數	43
Table 17 PTT電信各月影響準確率原因筆數	43
Table 18 PTT網路各月影響準確率原因筆數	43
Table 19 Mobile01電信各月影響回收率原因筆數	44
Table 20 Mobile01網路影響回收率原因筆數	44
Table 21 PTT電信各月影響回收率原因筆數	45
Table 22 PTT網路各月影響回收率原因筆數	45
Table 23 Mobile01電信假設句筆數	54
Table 24 PTT電信未加入預設Topic各月數據	55

[1]	J. A. Horrigan, "Online shopping," Pew Internet & American Life Project Report,2008.
[2]	K. Group, "Online consumer-generated reviews have significant impact on offline purchase behavior," 2007.
[3]	H. Zhang, Z. Yu, M. Xu, and Y. Shi, "Feature-level sentiment analysis for Chinese product reviews," 2011 pp. 135-140.
[4]	姚天昉, 程希文, 徐飞玉, 汉思, and 王睿, "文本意见挖掘综述," 中文信息学报, vol. 22, pp. 71-80, 2008.
[5]	D. McClosky, W. Che, M. Recasens, M. Wang, R. Socher, and C. D. Manning, "Stanford’s System for Parsing the English Web," 2012.
[6]	G. Qiu, B. Liu, J. Bu, and C. Chen, "Opinion Word Expansion and Target Extraction through Double Propagation," Computational Linguistics, vol. 37, pp. 9-27, 2011/03/01 2011.
[7]	簡立, "意見探勘系統設計," 淡江大學資訊工程研究所碩士論文, 2012.
[8]	楊盛帆, "以整合式規則來做網路論壇上的 3C 產品口碑分析," 元智大學資訊管理學系研究所碩士論文, 2009.
[9]	侯锋, 王传廷, and 李国辉, "网络意见挖掘, 摘要与检索研究综述," 计算机科学, vol. 36, pp. 15-19, 2009.
[10]	A.-M. Popescu and O. Etzioni, "Extracting product features and opinions from reviews," presented at the Proceedings of the conference on Human Language Technology and Empirical Methods in Natural Language Processing, Vancouver, British Columbia, Canada, 2005.
[11]	X. Ding, B. Liu, and P. S. Yu, "A holistic lexicon-based approach to opinion mining," presented at the Proceedings of the international conference on Web search and web data mining, Palo Alto, California, USA, 2008.
[12]	C. C. Yang and Y. C. Wong, "MINING CONSUMER OPINIONS FROM THE WEB," 2008.
[13]	N. Kobayashi, K. Inui, and Y. Matsumoto, "Opinion Mining from Web Documents: Extraction and Structurization," Information and Media Technologies, vol. 2, pp. 326-337, 2007.
[14]	N. Kobayashi, K. Inui, Y. Matsumoto, K. Tateishi, and T. Fukushima, "Collecting Evaluative Expressions for Opinion Extraction
Natural Language Processing – IJCNLP 2004." vol. 3248, K.-Y. Su, J. i. Tsujii, J.-H. Lee, and O. Kwong, Eds., ed: Springer Berlin / Heidelberg, 2005, pp. 596-605.
[15]	M. Hu and B. Liu, "Mining and summarizing customer reviews," presented at the Proceedings of the tenth ACM SIGKDD international conference on Knowledge discovery and data mining, Seattle, WA, USA, 2004.
[16]	S.-M. Kim and E. Hovy, "Determining the sentiment of opinions," presented at the Proceedings of the 20th international conference on Computational Linguistics, Geneva, Switzerland, 2004.
[17]	林偉揚, "應用種子詞彙延伸方式於BBS電影評論之口碑分析," 元智大學資訊管理學系研究所碩士論文, 2011.
[18]	Q. Su, X. Xu, H. Guo, Z. Guo, X. Wu, X. Zhang, B. Swen, and Z. Su, "Hidden sentiment association in chinese web opinion mining," 2008, pp. 959-968.
[19]	M. Fuketa, Y. Kadoya, E. Atlam, T. Kunikata, K. Morita, S. Kashiji, and J. I. Aoe, "A method of extracting and evaluating good and bad reputations for natural language expressions," International Journal of Information Technology and Decision Making, vol. 4, pp. 177-196, 2005.
[20]	V. Hatzivassiloglou and K. R. McKeown, "Predicting the semantic orientation of adjectives," presented at the Proceedings of the eighth conference on European chapter of the Association for Computational Linguistics, Madrid, Spain, 1997.
[21]	S. Morinaga, K. Yamanishi, K. Tateishi, and T. Fukushima, "Mining product reputations on the Web," presented at the Proceedings of the eighth ACM SIGKDD international conference on Knowledge discovery and data mining, Edmonton, Alberta, Canada, 2002.
[22]	批踢踢 (Ptt). Available: http://www.ptt.cc/index.html
[23]	Mobile01. Available: http://www.mobile01.com/
[24]	瞿怡正, "Mobile01和PTT兩個不同論壇相同面向習慣用語之探討-以電信寬頻為例," 淡江大學資訊工程研究所碩士論文, 2012.