網絡口碑社群使用者參與之訊息傳播行為研究-以Mobile 01為例

現今社會，資訊傳播快速，該如何有效控管資訊對於品牌造成的損害以及從中獲取效益，已然成為各個企業所面臨的機會與難題，其中根據不同產業及品牌又會有不同的情況所產生，這份論文主要便透過手機品牌來探討如何區分訊息擴散中不同的角色表現，以及了解不同品牌間影響者的形成模式，以此為目的，透過具有電子口碑傳播意向的意見交流平台Mobile01來建立六個社會網絡關係圖型.接著利用回應數目區分出具有冪次分配的影響者，透過社會網絡分析了解影響者的訊息傳播參與行為以及透過二元羅吉斯回歸區分出具有拓樸特徵的影響者與非影響者，以此了解不同手機品牌的傳播特性，以便進行管理。這個結果將暴露出影響者並不單單只決定於回應的篇數，還需考量回應的文章人氣程度，且根據不同品牌的特性需佔據不同的網絡位置，這篇論文主要的貢獻在於提供一個客觀且量化的方式，區分二種不同的訊息擴散的角色以此去探討不同的品牌訊息傳播的特性、角色位置及影響者形成的模式，且根據最後的結果發現到不同的品牌中影響者形成模式可分成回應導向、文章人氣導向、及綜合導向三種，透過分析不同的形成模式將可更有效的區分出正確的影響者來，且根據分析結果每個品牌中影響者所處的傳播環境都有所不同，管理方式上選擇不同的方法進行管理效果也會有相當大的差異，根據研究結果大致分為個別、群聚、整體性三種，透過不同的管理選擇，將使傳播會有擴散或收斂的效果，將更有利於企業進行管理。不同於其他論文，並不是利用問卷或是意見調查，而是透過抓取使用者在於網站上遺留下來的歷史評論、回應建立而成，屬於真實的第一手資料。

關鍵字

電子口碑；訊息擴散；社會網絡分析；社群媒體；二元羅吉斯回歸

並列摘要

This paper is focus on how to identify the roles of information diffusion behavior, and realize the forming mode of the influencers of different mobile brands. For this purpose, we will establish 6 social network models from the forum of Mobile01 which is one of the most famous EWOM platforms in Taiwan. And then through the quantities of response to filter the influencers of power low.Finally, we use social network analysis to realize the influencers’ information diffusion participate behavior and also use Binary Logistic Regression to identify the influencers and non-influencers of topological feature. This result will expose that how to become an influencer is not only decided by quantities of response but also by reviews popularity. Following the feature of different brand, influencers have dissimilar network position. This paper’s major contribution is providing an objective and quantified method to identify the information diffusion of different role. And then from this analysis, we can realize features of information diffusion, roles position, and the forming mode of the influencers about different brands. Finally we find the forming mode of the influencers can be classified by three methods: response orientation, popularity orientation and complex orientation. Through this analysis, we can use the different forming mode to accurately screen out the influencers. And we find the different mobile brands have dissimilar information diffusion environment which could be divided into individual, group and overall. Then we can use this three categories to make message management easier. In order to obtain real and first-hand data which also different from other papers, rather than questionnaires or opinion polls, we use the Web crawler to grab users’ historical comments which left on the sites. So it will be more valuable for future research.