網路線上社群已成為現代人生活的一部分，透過Twitter、Facebook、Line等網路社群，人們便可隨時隨地分享心情、交換訊息。為有效管理社群，社群管理者經常需藉由高影響力成員的協助，尋求他們的合作。此外，管理當局對於社群討論方向也需隨時掌握，以免造成社群的紛爭或會員流失。有鑒於此，本研究將參考生物群聚(animal flocking)之行動概念，建立社群動向之評估模型，並藉以標示社群中的影響力成員，以進行有效的社群管理。首先，我們先為討論區文章建立類型，以做為定義社群動向的依據。之後，我們再將社群的討論方向定義為社群中各類型文章佔有的比例。藉由統計各週間討論方向之相似度，從中篩選造成改變之主要成員，並評估這些成員對討論方向改變的重要性。為驗證提出方法之有效性，本研究蒐集PTT Stock與iOS討論版三年共156週實際發文資料進行實驗。結果顯示在特殊時間點，社群的討論方向確實會發生明顯改變，證明提出之方法確能反映社群動向。此外，我們也比較三種不同方法在社群動向改變時之反映程度。根據這些變化，我們篩選出社群中的影響力成員，並與傳統社群網路分析指標比較。結果顯示本研究提出方法與傳統指標有明顯差異，能有效標示實際造成社群動向改變之成員。上述成果說明本研究提出之方法具有實用性，可提供管理者更準確之決策依據。

Online communities have become a part of life. People can share their feelings or exchange messages anytime, anywhere through the online community of (e.g., Twitter, Facebook, and Line). In order to effectively manage the community, community managers often need to seek cooperation with the help of high-impact members. In addition, the management needs to keep track of the direction of community discussions, so as to avoid community disputes or loss of membership. Therefore, this study will use the concept of animal flocking to establish an assessment model of community trends and use it to identify influence members in the community for effective community management. First, we create types for discussion forum articles as a basis for defining community trends. After that, we define the discussion direction of the community as the proportion of each type of article in the community. By investigating the similarities of the discussion directions in each week, the main members of the change are screened and the importance of these members to the direction of the discussion is assessed. In order to verify the validity of the proposed method, this study collected PTT Stock and iOS discussion forum for 156 weeks of actual publication data for experiment. The results show that at a particular point in time, the discussion direction of the community will change significantly, proving that the proposed method does reflect community trends. In addition, we also compare the extent to which three different approaches reflect changes in community dynamics. Based on these changes, we screen out the influence members in the community and compare them with traditional social network analysis indicators. The results show that the proposed method is significantly different from the traditional indicators, and it can identify the members that actually cause changes in the community movement more effectively. The above results show that the method proposed in this study is helpful for analyzing the development trend of the community and providing more accurate decision-making basis for community managers.

目錄
第一章 緒論	1
第二章 背景知識與相關技術介紹	4
2.1 生物群聚模型(Animal Flocking Model)	4
2.2 各種效能評量指標	5
2.3 社群網路分析指標	6
第三章 找尋改變社群動向之影響力成員	9
3.1	建立文章類型	9
3.2	討論區文章分類	10
3.3	評估社群動向之改變	14
3.3.1 建立社群動向之模型	14
3.3.2評估社群動向改變之流程	16
3.4 找尋改變討論動向的影響力成員	19
第四章 實驗結果	21
4.1 資料蒐集	21
4.2 社群討論方向之變化	22
4.3 找尋社群影響力成員	24
第五章 結論與未來工作	29
參考文獻	31
附錄: 各討論區之各種討論動向表示法之比較(2016-2019)	34

 
表目錄
表3-1 不具特別意義之用語	10

表3-2 PTT IOS討論區選用各週前N名關鍵詞之聯集大小(共156週)	10

表3-3: 以社群發文類型比率表示社群動向之範例	15

表3-4:  PTT八卦版在10週期間各類型文章之分布比例	16

表4-1:以不同方式評估之討論區前20名影響力成員(PTT STOCK版, 140-141週)	25

表4-2: 各種方法找出之前五名影響力成員之比較 (PTT STOCK版)	26

表4-3:以不同方式評估之討論區前20名影響力成員(PTT IOS版, 120-121週)	27

表4-4: 各種方法找出之前五名影響力成員之比較 (PTT IOS版)	28


 
圖目錄
圖 2-1生物群聚模型: 生物移動方向受鄰近成員影響	4

圖 3-1: PPT IOS版文章分群之GAP STATISTIC變化圖	11

圖 3-2: PPT STOCK版文章分群之GAP STATISTIC變化圖	11

圖3-3: 根據人工方式產生之文章類型進行文章分類之虛擬碼	13

圖3-4: PTT八卦板在資料蒐集期間(11週)，各週之間討論動向之變化	16

圖3-5: 評估社群討論動向改變之流程圖	18

圖 3-6 根據造成之社群討論動向差異值找尋影響力成員之虛擬碼	20

圖4-1 使用PTT CRAWLER下載討論區資料之運作流程	21

圖4-2 三種討論動向表示方式之週間相似度變化比較(PTT STOCK版)	23

圖4-3 三種討論動向表示方式之週間相似度變化比較(PTT IOS版)	24

圖A-1  2016年5月-2017年5月PTT STOCK版之動態變化	34

圖A-2  2017年5月-2018年5月PTT STOCK版之動態變化	34

圖A-3  2018年5月-2019年5月PTT STOCK版之動態變化	35

圖A-4  2016年5月-2017年5月PTT IOS版之動態變化	35

圖A-5  2017年5月-2018年5月PTT IOS版之動態變化	36

圖A-6  2018年5月-2019年5月PTT IOS版之動態變化	37

1.Cheng, H.-T., et al., "Wide & Deep Learning for Recommender Systems," in ACM DLRS ’16, Sep. 15, 2016, Boston, MA, USA.

2.Chin, A., et al., "Connecting People through Physical Proximity and Physical Resources at a Conference," ACM Transactions on Intelligent Systems and Technology, Vol. 4, No. 3, Article 50, June 2013, pp. 50-70.

3.Cohen, J., "Statistical Power Analysis for the Behavioral Sciences", Chapter 3, 1977, pp.77-81.

4.Covington, P., Jay Adams, Emre Sargin, "Deep Neural Networks for YouTube Recommendations," in Proceedings of RecSys ’16 September 15-19, 2016, Boston , MA, USA. pp. 191-198.

5.Cruz, J. D. Bothorel, Ce´clie, Poulet, F., "University of Rennes I - IRISA Community Detection and Visualization in Social Networks: Integrating Structural and Semantic Information," ACM Transactions on Intelligent Systems and Technology, Vol. 5, No. 1, Article 11, December 2013. pp. 11-36.

6.Franks, H., Griffiths, N. and Anand, S. S., "Learning influence in complex social networks." Proceedings of the 2013 international conference on Autonomous agents and multi-agent systems.

7.Franks, H., Griffiths, N. and Anand, S. S., "Learning agent influence in MAS with complex social networks," Auton Agent Multi-Agent Syst (2014) 28:836-866. 

8.Genter, K., and P. Stone, "Influencing a flock via ad hoc teamwork," In Ninth International Conference on Swarm Intelligence (ANTS’14), September 2014.

9.Genter, K., S. Zhang, and P. Stone. "Determining placements of influencing agents in a flock," in Proceedings of AAMAS’15, May 2015.

10.Genter, K. and P. Stone. Ad hoc teamwork behaviors for influencing a flock. Acta Polytechnica, 56(1):18–26, 2016.

11.Genter, K., S. Zhang, and P. Stone, "Adding Influencing Agents to a Flock," in Proceedings of the 15th International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2016), 615-623.

12.Jiang, W., et al., "A New Evaluation Algorithm for the Influence of User in Social Network," China Communication, Feb. 2016. pp. 200-206.

13.Kim, Y., "Convolutional Neural Networks for Sentence Classification," Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 1746–1751, October 25-29, 2014, Doha, Qatar.

14.Liu., Y., et al., "Modeling What Friendship Patterns on Facebook Reveal About Personality and Social Capital," ACM Transactions on Computer-Human Interaction, Vol. 21, No. 3, Article 17, May 2014. pp. 16-36.

15.Lynch, C., et al., "Images Don’t Lie: Transferring Deep Visual Semantic Features to Large-Scale Multimodal Learning to Rank," in Proceedings of KDD ’16, Aug. 13-17, 2016, San Francisco, CA, USA. pp. 540-548.

16.Reynolds, C. W., "Flocks, Herds, and Schools: A Distributed Behavioral Model," Computer Graphics, 21(4), July 1987, pp. 25-34.

17.Tibshirani, Robert, Guenther Walther, and Trevor Hastie. "Estimating the number of clusters in a data set via the gap statistic." Journal of the Royal Statistical Society: Series B (Statistical Methodology) 63.2 (2001): 411-423.

18.Shafiq, M. Z., et al., "Identifying Leaders and Followers in Online Social Networks," IEEE Journal on Selected Areas in Communications/Supplement, Vol. 31, No. 9, Sep. 2013. pp. 618-628.

19.Sharma, M. K., et al., "Abnormality Detecting Deep Belief Network," in Proceedings of AICTC ’16, August 12-13, 2016, Bikaner, India.

20.Socher, R., et al., "Recursive Deep Models for Semantic Compositionality Over a Sentiment Treebank," In Proceedings of the 2013 Conference on Empirical Methods in Natural Language Processing, pp 1631-1642. 2013.

21.Su, H., Wang., X., and Lin, Z., "Flocking of Multi-Agents With a Virtual Leader," IEEE Trans. on Automatic Control, VOL. 54, NO. 2, Feb. 2009, pp. 293-307.

22.Subbian, K., et al., "Mining Influencers Using Information Flows in Social Streams," ACM Transactions on Knowledge Discovery from Data, Vol. 10, No. 3, Article 26, January 2016.

23.Tai, Kai-Sheng, Richard Socher, Christopher D. Manning, "Improved Semantic Representations From Tree-Structured Long Short-Term Memory Networks," Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing, pages 1556–1566.

24.Tang., X. and Yang, C. C., "Ranking User Influence in Healthcare Social Media," ACM Transactions on Intelligent Systems and Technology, Vol. 3, No. 4, Article 73, September 2012. pp. 73-93.

25.Tavakolifard, Mozhgan, and Kevin C. Almeroth. "Social computing: an intersection of recommender systems, trust/reputation systems, and social networks." Network, IEEE 26.4 (2012): 53-58.

26.Wang, W., et al., "Deep Learning at Scale and at Ease," ACM Trans. Multimedia Comput. Commun. Appl., Vol. 12, No. 4s, Article 69, Nov. 2016, p. 69-93.

27.Yuan, Z., et al., "Latent Feature Learning in Social Media Network," in ACM MM’13, October 21–25, 2013, Barcelona, Spain. pp. 253-262.

28.Wang, C., Chen, W. and Wang, Y., "Scalable Influence maximization for independent cascade model in large-scale social networks," Data Mining and Knowledge Discovery, Nov. 2012, Vol. 25, Issue 3, pp 545–576.

29.Zhang, H., et al., "Misinformation in Online Social Networks: Detect Them All with a Limited Budget," ACM Transactions on Information Systems, Vol. 34, No. 3, Article 18, April 2016. pp. 18-42.

30.Yu-Ting Wen(2014) "Exploring Social Influence on Location-Based Social Networks," National Chiao Tung University the Degree of  Master in Computer Science , Hsinchu, Taiwan, Republic of China

31.Fernando Henrique Calderon Alvarado(2015), "Influence Value: Quantifying Topic Influence in Social Media," National Tsing Hua University the Degree of Master in Institute of Information Systems and Applications , Hsinchu, Taiwan, Republic of China

32.Chun-Jie Chen(2017), "Mobility-aware Influence Maximization in Location-based Social Networks," National Chiao Tung University the Degree of  Master in Computer Science , Hsinchu, Taiwan, Republic of China