|Abstract: ||社群網站為社交、分享、組織管理、及金融交易等活動提供有效的平台。身為最 大的社群網站，臉書(Facebook)擁有八億多名使用者及九億多個互動項目﹝包括個 人、團體、事件、與社群網頁﹞1。主要對手如Google、Microsoft 也不斷在發展自己 的社群網站。 社群網站在商業上日益重要，資訊擴散管理也因此成為必需。在電子商務的環境 中，顧客無法體驗產品、對供應商的說詞缺乏信心，口碑效果更為重要。因而網路口 碑傳播可為潛在顧客提供可靠的資訊。 本計畫將探討兩個社群網站資訊擴散的重要議題： 1. 預測貼文的影響力：社群網站的某些使用者可主導它人對產品或服務的印象、評價、 及購買決策。準確及時的找出他們是發展資訊擴散策略的基礎。此外，資訊內容特 徵，如語意、媒體類型也同樣會造成影響。本研究整合這兩個議題，提出以資料探 勘技術為基礎的架構，以辨別有影響力的人並預測一篇貼文一段時間後的影響力。 2. 使用者散播網路口碑的方式與理由：藉由好奇理論及學習方面的文獻，我們探討社 群網站的口碑接收者如何轉變為口碑傳播者，及促成其傳遞行為的因素。本研究提 出理論架構以證實並分析社群網站使用者傳遞行為的決定因子。我們將在臉書上收 集資料以檢測研究模型，並討論其理論、實務意涵及未來研究方向。|
Social Networking Sites (SNSs) provide an effective online platform for organizing human activities, including social contacts, idea and value sharing, managing membership of organizations, financial exchanges, etc. As the largest social networking site, Facebook has more than 800 million active users and more than 900 million objects that people interact with (pages, groups, events and community pages)1 by November 2011. Other major players, such as Google and Microsoft have been aggressively developing their own sites. As SNSs are more and more important for business, managing the diffusion process of information is essential. In the e-commerce environment, word-of-mouth effects are particularly important because consumers are unable to physically examine products and lack of trust in advertisements and reviews published by vendors. Consequently, electronic word-of-mouth (eWOM) communication may provide more dependable information to potential customers. In this proposed project, we aim to study two key issues associated with information diffusion on SNSs. The first issue is prediction of influence of posts and the second is on the reasons and methods people spread eWOM. 1. Predicting the Influence of Posts: In developing information diffusion strategies, a fundamental task is accurate and timely identification of individuals or nodes on SNSs, which strongly influence others in forming opinions, evaluations, and purchasing decisions on a product or services. In addition, the characteristics of content, such as sentiment and media type, also impact the results. This study integrates the two issues and proposes a framework based on data mining techniques for identifying influencers and predicting the influence of a post after a period of time. 2. Reasons and methods people spread eWOM: We plan to study and explore how eWOM receivers transform to eWOM senders on SNSs and what factors trigger pass-along behaviors on SNSs from theory of curiosity and the literature on learning. This study proposes a theoretical framework, which is to be validated, to investigate the factors that determine a sender’s pass-along behaviors on SNSs. Data will be collected from users of Facebook to test the research model. Implications for theory and practice and suggestions for future research will be discussed.