當社群發文者於社群討論環境中發表各式各樣的文章時，若此些社群發文者所發表
之社群發文能得到社群閱讀者高度的關注與迴響，則此些社群發文者往往被視為具有社
群吸引力的社群領袖。但隨著每次社群發文內容的變化，並非每次社群發文者所發表之
發文都能引起閱讀者的關注與迴響，亦即並非每次社群發文者的社群發文都具備社群吸
引力。
為協助社群發文者了解具備何種條件之社群發文才能吸引閱讀者之關注，本研究乃
發展一套「以社群發文內容為基礎之社群吸引力解析」模式。為建構此模式，首先，本
研究於前置階段透過社群討論環境（如Facebook）蒐集各社群發文者所發表之社群發文，
並分析各社群發文中可能影響社群吸引力之潛在因子，最後歸納共五大類別（即「潛在
因子接續於特定英文字母組合之後」、「潛在因子以特殊符號或特定詞彙開頭」、「潛在因
子直接以特殊詞彙表達」、「直接性系統性因子」、「間接性系統性因子」等類別）、11 項
之潛在因子（即外部連結數、標註 Hashtag、同伴打卡、地標打卡、關鍵詞彙、文字數、
照片數、留言數、分享數、標點符號數、標點符號數等潛在因子）；本研究並依此些潛
在因子之表達方式建立詞庫，以利後續階段進行社群發文內容之潛在因子擷取。之後，
本研究以前置階段為基礎發展「以社群發文內容為基礎之社群吸引力解析」方法論，此
方法論可區分為「潛在因子擷取」、「具社群吸引力之社群發文判定」及「關鍵潛在因子
判定」等三大階段。其中，「潛在因子擷取」階段將前置階段所建置之詞庫與各社群發
文內容進行比對，以擷取各社群發文內容之潛在因子。其次，「具社群吸引力之社群發
文判定」階段乃採用分群法依各社群發文對應之讚數將所有社群發文進行分群，並根據
分群之結果判定具社群吸引力之社群發文。最後，「關鍵潛在因子判定」階段乃根據前
一階段之社群發文分群結果，進行各別潛在因子於兩群集中各自平均值之假設檢定，以
判斷各別潛在因子是否具顯著差異，進而釐清對社群吸引力具關鍵影響性之關鍵因子；
而本研究於此階段判定三項關鍵因子對社群吸引力具關鍵影響性，此三項關鍵因子分別
為「照片」、「標註（Hashtag）」、「表情符號」等。透過運用此三項關鍵因子並搭
配個人魅力與人格特質，社群發文者得以發表足夠吸引閱讀者關注與迴響之社群發文。

As the social media publisher issues a post on the social media, they get a high degree of concern and his/her post might grab much social media attraction and great responses from the social media readers. The publishers with many responses from the public can be regarded as social media leaders. However, with the
changes in the contents of posts, the publisher might not always grab the reader attention and great responses;
that is, the social media publisher does not have a great attraction all the time.
In order to reveal the factors that affect the attraction of the social media post, this research collected a great number of posts and analyzed the potential factors which might influence the attraction and responses from the social media readers. Furthermore, this research develops a model for attraction analysis of social media posts.
The proposed model can be used to extract the contents of the potential factors from the posts and analyze the expression structure from posts. After that, the model utilizes the k-means method to identify the social media posts with great attraction and applies the hypothesis test to determine the key factors for reader attraction. On the basis of key factors and personal charisma, the social media publisher might issue the attractive posts to catch the social media reader attention and great responses.

摘要 ........................................................................................................................................... II
ABSTRACT ........................................................................................................................... III
目錄 .......................................................................................................................................... IV
第一章、研究背景 .................................................................................................................... 1
1.1 研究動機與目的 .............................................................................................................. 1
1.2 研究流程 .......................................................................................................................... 3
第二章、文獻回顧 .................................................................................................................... 6
2.1 社群發文特徵擷取 ......................................................................................................... 6
2.1.1 以監督式方法擷取社群發文特徵 ........................................................................... 6
2.1.2 以非監督式方法擷取社群發文特徵 ...................................................................... 8
2.1.3 以混合式方法擷取社群發文特徵 ......................................................................... 12
2.2 社群內容價值推論 ....................................................................................................... 14
2.2.1 以機器學習演算法進行社群內容價值推論 ......................................................... 14
2.2.2 以統計方法進行社群內容價值推論 ..................................................................... 20
2.2.3 以語言處理方法進行社群內容價值推論 ............................................................. 23
2.3 社群網站領袖與專家推論 ........................................................................................... 27
2.3.1 以社群發文為基礎之社群網站領袖與專家推論 ................................................. 27
2.3.2 以社交圈為基礎之社群網站意見領袖或領域專家推論 ..................................... 31
第三章、以社群發文內容為基礎之社群吸引力解析模式 .................................................. 35
3.1 社群發文內容解析 ....................................................................................................... 37
3.1.1 社群發文所具備之潛在因子釐清 ........................................................................ 37
3.1.2 潛在因子表達結構歸納 ......................................................................................... 40
3.2 潛在因子擷取 ................................................................................................................ 46
3.3 具社群吸引力之社群發文判定 .................................................................................... 54
3.4 關鍵潛在因子判定 ........................................................................................................ 59
3.5 小結 ................................................................................................................................ 63

V
第四章、模式績效驗證與分析 .............................................................................................. 64
4.1 模式驗證方式說明 ........................................................................................................ 64
4.2 模式驗證結果分析 ........................................................................................................ 68
4.2.1 第一階段驗證結果分析 ......................................................................................... 69
4.2.2 第二階段驗證結果分析 ......................................................................................... 72
第五章、結論與未來展望 ...................................................................................................... 75
5.1 論文總結 ....................................................................................................................... 75
5.2 未來展望 ....................................................................................................................... 78
參考文獻 .................................................................................................................................. 79
附錄 A、社群發文蒐集及解析 ............................................................................................. 83
附錄 B、驗證資料蒐集 ........................................................................................................ 107
附錄 C、模式於第二階段各週期之績效驗證結果 ............................................................ 403

Shady, E., Muhammad, I., Carlos, C., Fernando, D. and Patrick, M., 2013, “Extraction
Information Nuggets From Disaster-Related Messages in Social Media,” Proceeding of
the 10th ISCRAM Conference, pp.1-10.
2. Akaichi, J., 2013, “Social Networks’ Facebook’ Statutes Updates Mining for Sentiment
Classification,” Social Computing, pp. 886-891.
3. Muhlhauser, M., Gurevych, I. and Weimer, M., 2007, “Automatically Assessing the Post
Quality in Online Discussions on Software,” Proceedings of the 45th Annual Meeting of
the Association for Computational Linguistics Companion Volume Proceedings of the
Demo and Poster Sessions, pp. 125-128.
4. Funda, K.S., Mor, N. and Nicholas, D., 2010, “Diamonds in the Rough: Social Media
Visual Analytics for Journalistic Inquiry,” Analytics Science and Technology, Vol. 29, pp.
115-122.
5. Farahmandnia, F., Omidvar, A. and Kardan, A., 2011, “Expert Finding on Social
Network with Link Analysis Approach,” Electrical Engineering, pp. 1-6.
6. Jehan W., Siqi Z., Lin Z. and Venu V., 2010, “Analyzing Twitter for Social TV:
Sentiment Extraction for Sports,” FutureTV Workshop, Vol. 720, pp. 140-145.
7. Pomerantz, J. and Shah, C., 2010, “Evaluating and Predicting Answer Quality in
Community QA,” Proceedings of the 33rd International ACM SIGIR Conference on
Research and Development in Information Retrieval, pp. 411-418.
8. Nebhi, K., 2012, "Ontology-Based Information Extraction from Twitter," Proceedings of
the Workshop on Information Extraction and Entity Analytics on Social Media Data, pp.
17-22.
9. Bernstein, M. S., André, P. and Luther, K., 2012, “Who Gives A Tweet? Evaluating
Microblog Content Value,” Proceedings of the ACM 2012 Conference on Computer
Supported Cooperative Work, pp. 471-474.

80
10. Yeh, P. W. and Hung, C., 2014, “Identification of Opinion Leaders Using Text Mining
Technique in Virtual Community,” Proceedings of the 1st Symposium on Information
Management and Big Data, Vol. 1318, pp. 8-13.
11. Jiang, L., Duan, Y., Qin, T., Zhou, M. and Shum, H. Y., 2010, “An Empirical Study on
Learning to Rank of Tweets,” Proceedings of the 23rd International Conference on
Computational Linguistics, pp. 295-303.
12. han, M. K., Li, C. T., Jheng, S. H. and Chou, K. C., 2016, “Exploiting Concept Drift to
Predict Popularity of Social Multimedia in Microblogs,” Information Sciences, Vol. 339,
pp. 310-331
13. Leung, K. W., Vosecky, J. and Ng, W., 2012, “Searching for Quality Microblog Posts:
Filtering and Ranking Based on Content Analysis and Implicit Links,” DASFAA, Vol.
7238, pp. 397-413.
14. Efron, M., 2010, “Hashtag Retrieval in a Microblogging Environment,” Proceedings of
the 33rd International ACM SIGIR Conference on Research and Development in
Information Retrieval, pp. 787-788.
15. Jiajun, Y. and Ying, L., 2015, “Improving Ranking-based Recommendation by Social
Information and Negative Similarity,” Procedia Computer Science, Vol. 55, pp. 732-740
16. Wang, W. M., Tsui, E., Cai, L., Cheung, C. F. and Lee, W. B., 2014, "Knowledge-Based
Extraction of Intellectual Capital-Related Information from Unstructured Data," Expert
Systems with Applications, Vol. 41, pp. 1315-1325.
17. Wayne, X. Z., Jing J., Jing, H., Yang, S., Palakorn, A., Ee, P. L. and Xiaoming, L., 2011,
“Topical Keyphrase Extraction from Twitter,” Computational Linguistics: Human
Language Technologies, pp. 379-388.
18. Ercan, G. and Cicekli, I., 2007, “Using Lexical Chains for Keyword Extraction,”
Information Processing and Management, Vol. 43, pp. 1705-1714.
19. Lloret, E. and Palomar, M., 2013, "Towards Automatic Tweet Generation: A
Comparative Study from the Text Summarization Perspective in the Journalism Genre,"
Expert Systems with Applications, Vol. 40, pp. 6624-6630.

81
20. Leung, K. W., Vosecky, J. and Ng, W., 2012, “Searching for Quality Microblog Posts:
Filtering and Ranking Based on Content Analysis and Implicit Links,” DASFAA, Vol.
7238, pp. 397-413.
21. Muhlhauser, M., Gurevych, I. and Weimer, M., 2007, “Automatically Assessing the Post
Quality in Online Discussions on Software,” Proceedings of the 45th Annual Meeting of
the Association for Computational Linguistics Companion Volume Proceedings of the
Demo and Poster Sessions, pp. 125-128.
22. Pomerantz, J. and Shah, C., 2010, “Evaluating and Predicting Answer Quality in
Community QA,” Proceedings of the 33rd International ACM SIGIR Conference on
Research and Development in Information Retrieval, pp. 411-418.
23. Wayne, X. Z., Jing J., Jing, H., Yang, S., Palakorn, A., Ee, P. L. and Xiaoming, L., 2011,
“Topical Keyphrase Extraction from Twitter,” Computational Linguistics: Human
Language Technologies, pp. 379-388.
24. Bernstein, M. S., André, P. and Luther, K., 2012, “Who Gives A Tweet? Evaluating
Microblog Content Value,” Proceedings of the ACM 2012 Conference on Computer
Supported Cooperative Work, pp. 471-474.
25. Shan, M. K., Li, C. T., Jheng, S. H. and Chou, K. C., 2016, “Exploiting Concept Drift to
Predict Popularity of Social Multimedia in Microblogs,” Information Sciences, Vol. 339,
pp. 310-331
26. Jiajun, Y. and Ying, L., 2015, “Improving Ranking-based Recommendation by Social
Information and Negative Similarity,” Procedia Computer Science, Vol. 55, pp. 732-740
27. Shady, E., Muhammad, I., Carlos, C., Fernando, D. and Patrick, M., 2013, “Extraction
Information Nuggets From Disaster-Related Messages in Social Media,” Proceeding of
the 10th ISCRAM Conference, pp.1-10.
28. Gerster, D., Alonso, O., Carson, C., Ji, X. and Nabar, S., 2010, “Detecting Uninteresting
Content in Text Streams,” Proceedings of SIGIR CSE Workshop, pp. 39-42.

82
29. Benyoucef, M., Kem, Z. and Sesia, Z., 2016, “Building Brand Loyalty in Social
Commerce: The Case of Brand Microblogs,” Electronic Commerce Research and
Applications, Vol. 15, pp. 14-25.
30. Liang, F., Qiang, R. and Yang, J., 2013, “Exploiting Ranking Factorization Machines for
Microblog Retrieval,” Information & Knowledge Management, pp. 1783-1788.
31. Tsagkias, M., Massoudi, K., Rijke, M. D. and Weerkamp, W., 2011, “Incorporating
Query Expansion and Quality Indicators in Searching Microblog Posts,” Advances in
Information Retrieval, Vol. 6611, pp. 362-367.
32. Bernhard, D., Zhu, Z. and Gurevych, I., 2009, “A Multi-Dimensional Model for
Assessing the Quality of Answers in Social Q&A Sites,” Technical Report, Vol. 156, pp.
1-13.
33. Weerkamp, W., Meij, E. and Rijke, M. D., 2012, “Adding Semantics to Microblog Posts,”
Web Search and Data Mining, pp. 563-572.
34. Ashour, H., Wanas, N., Saban, M. E. and Ammar, W., 2008, “Automatic Scoring of
Online Discussion Posts,” Proceedings of the 2nd ACM Workshop on Information
Credibility on the Web, pp. 19-26.
35. Chung, Y. and Akter, R., 2013, “An Evolutionary Approach for Document Clustering,”
IERI Procedia, Vol. 4, pp. 370-375.3
36. Farahmandnia, F., Omidvar, A. and Kardan, A., 2011, “Expert Finding on Social
Network with Link Analysis Approach,” Electrical Engineering, pp. 1-6.
37. Yeh, P. W. and Hung, C., 2014, “Identification of Opinion Leaders Using Text Mining
Technique in Virtual Community,” Proceedings of the 1st Symposium on Information
Management and Big Data, Vol. 1318, pp. 8-13.