以視覺化系統尋找社群媒體中的意見領袖

Design of a Visualization System for Finding Opinion Leaders in Social Media

楊恩加(Yang En-Jea)

指導教授 ： 李蔡彥

國立政治大學/理學院/資訊科學系/碩士(2019年)

https://doi.org/10.6814/THE.NCCU.CS.005.2019.B02

摘要

意見領袖（Opinion Leader），是人類社群在討論議題時自然形成的特色人物。從過往研究中，我們得知他們能有效並快速地將新消息傳達給大眾，並引導大眾的思考方向和結論。故此，我們想要深入地探討新聞事件的傳遞模式和樣貌，找出其中的意見領袖，和他們與大眾的連結關係。過往在人類行為學的領域上，已大致分出判斷意見領袖的十種方式。然而若直接套用在社群媒體中，則會因其數位化型態，互動方式的不同設定而無法切合應用。在本研究中，我們以在推特（Twitter）蒐集的三個主題事件（2015復興航空空難、2015亞洲投資銀行、2016台灣總統選舉）進行測試。藉由視覺化圖表、數據統計幫助研究者了解此主題事件的資料樣貌，再藉由數種找出意見領袖的方法，列舉出不同的候選人，並將其定位成重要連結者，以供研究者分析了解。為驗證本系統的可用性，我們邀請了五位受試者，透過操作教學、引導式任務讓受試者學習如何使用系統，接著讓受試者擇一事件找出三位以上的意見領袖，並透過問卷與訪談來探討系統的優缺點。實驗結果照示，系統在推特事件中能輔助研究者找出意見領袖。透過其視覺化輔助和資訊彙整，使用者能快速找到意見領袖，並以其在事件中與大眾的互動而更認識他們。

關鍵字

視覺化 ； 意見領袖 ； 推特 ； 社群媒體

並列摘要

Opinion leaders usually are formed naturally when people discuss issues. From the literature, we know that they can convey information to the public quickly, and lead the people to think and draw conclusions. Therefore, in this research we would like to explore how news events propagate in social media in order to find the opinion leaders and understand their relationship between the people. In the literature, there are ten ways to distinguish opinion leaders through the study of human behaviors. However, not all of them can be applied to social media directly because of the differences between real world and social media.
In our research, we have used three datasets collected from Twitter as case studies. The topics of these datasets include TransAsia Airways Flight 235, Asian Infrastructure Investment Bank and 2016 Taiwan presidential election. We have developed a visual analysis system to help researchers grasp the datasets in a broad sense, and then find candidates for opinion leaders by filtering the data with several indexes. The opinion leaders are then identified by further information such as the social network relations. To verify the usefulness of our system, we have invited five subjects to participate in a series of tasks to find three opinion leaders in a given event. The subjects are asked to evaluate the system and give textual feedbacks about their experiences. The experimental results reveal that through the visualization functions and data consolidation, this system can effectively help researchers find opinion leaders and know their roles in social media.

並列關鍵字

Visualization ； Opinion leader ； Twitter ； Social media

參考文獻

[1] W. contributors. "Web 2.0," 7 December 2018; https://en.wikipedia.org/w/index.php?title=Web_2.0&oldid=871056182.

[2] 1mayamaria, "Traditional Media Vs New Media," 2015.

[3] D. A. b. Marzuki, "Traditional Media Vs New Media," 2012.

[4] 食來運轉. "傳統媒體VS自媒體？," 7 December 2018; https://kknews.cc/zh-tw/other/o89pgq.html.

[5] W. contributors. "Two-step flow of communication," 7 December 2018; https://en.wikipedia.org/w/index.php?title=Two-step_flow_of_communication&oldid=849913478.