社群網路分析為一個能在不同情境下蒐集、分析及呈現社群間關係的方法。近年來由於社群網路服務的快速崛起及普及引領出新的網路消費族群，促使企業需要針對網際網路使用者，採用新的市場策略來發展及支援自身品牌、產品與服務在網路上的行銷。此外，隨著社群網站和微部落格的興起，例如：Facebook、Digg及Twitter，大量使用者在這些網站上建立個人資料檔、分享生活經驗與個人興趣及建立主題社群，也造成人們的互動模式與消費行為改變，更使得許多計算機學家、行銷市場專家及社會科學學者對這些社群平台上的使用者訊息如何透過社群網路對其它用戶產生影響，及其資訊如何散播的現象產生極大的興趣。而其中又以發現與確認社群中每個使用者的影響力大小，以支援企業在市場行銷上的使用，來達到理想的行銷成果為最熱門之研究主題。然而在過往研究中，大量的學者著重於利用統計及機率模組的方式來評估使用者訊息在社群網路中資訊散播的速度及涵蓋範圍，較缺乏從資料探勘的層面來探討使用者過往訊息的傳播行為；此外，亦未從較抽象化的思維來觀察使用者所屬社群之間的訊息傳播現象。因此本研究為提供一個有系統性之多階層訊息傳播規則評估方式，將依據如下步驟，漸次的探討社群網路中不同社群活動之訊息傳播行為。第一、設計一個能夠支援時序性查詢及多階層探勘之演算法，以有效挖掘及查詢不同細緻程度之訊息傳播行為。第二、針對社群網路中的多種不同主題性質之社群活動，設計一個分類方法來建構出該社群網路之概念化階層，以支援上述演算法來分析多階層社群活動。本研究期望結合社群網路分析及資料探勘方法來建立具高彈性化之多階層社群網路分析方法，以供不同主題領域之專家能夠從最細微之使用者訊息傳播行為，到最巨觀之社群間傳播活動皆能有效觀察分析，以因應Web 2.0 興起後快速演變之網路使用者行為模式。Social network analysis is a method for collecting, analyzing, and displaying community relationships under various scenarios. The growing pervasiveness of social network services has generated new customer groups enabling enterprises to adopt new marketing strategies for developing, marketing and supporting internet products and services. Well-known social networking websites such as Facebook, Twitter, and Digg allow users to construct personal profiles, share interesting information, and build relationships within a community. Interaction on social websites is rapidly affecting people’s social behaviors and habits. From our comprehensive survey of the literature on social networks showed that understanding the impact of communities in social networks requires complex computations. Furthermore, the few studies performed until now have focused on designing efficient and flexible platforms for mining the answers to questions in social networks to support social influence and relationship analysis, such as “What are the top-K direct influencing sources by a certain user or community?”, “Who is affected by a specific user or community at a given time?”, or even “What top-K users or communities generate the largest number of distinct paths during message propagation?”. This study therefore designed a flexible data mining algorithm for hierarchical analysis of the influence of social networks on emerging Web 2.0 technologies in order to analyze how this internet business model evolved.The proposed method has two major components.(1)A generalized suffix tree structure for analyzing information propagation patterns in communities with varying granularity.(2)A classification module for classifying social activities based on user interests. This work also designs a query interface for mining propagation paths and for discovering social network nodes that efficiently support enterprises in managing and analyzing social information. We look forward that the proposed mechanism has potential applications as a flexible social network platform for summarizing and analyzing network user behavior.
