網路廣告與虛擬社群使用者的分群機制之探討

A study on web advertising and the clustering mechanism for virtual community users

蔡佳晏(Chia-Yen Tsai)

指導教授 ： 游耿能

國立臺中科技大學/設計學群/多媒體設計系碩士班/碩士(2010年)

摘要

近年來，網路廣告隨著以Web2.0資訊網路為主的媒體快速發展，不僅廣告量大幅成長，也發展了豐富的廣告模式。其中，以網路廣告結合虛擬社群的行銷方式，演進為新型態的網路廣告機制，延伸至虛擬寵物社群平台。然而現今虛擬寵物社群平台上，利用網路廣告媒合使用者與虛擬寵物互動中所隱含的興趣偏好方面仍有不足之處。因此本研究觀察現有虛擬寵物社群平台，探討具代表性的平台之個案背景與服務功能，進一步分析會員註冊流程的項目與虛擬寵物設計，並整理各虛擬社群的網路廣告分類方式，以做為本研究建構廣告分類之依據。據此，本研究實作一個虛擬社群平台，設計以透過使用者在註冊會員的過程中，擷取使用者的偏好項目，並運用SVM(支援向量機)的分群技術，提供將使用者喜好結合網路廣告播放之媒合。本研究使用公開的Libsvm工具，並在分群的有效性上設計了一些實測。結果顯示，對於使用者在註冊流程後，系統以三類或六類的網路廣告分類，對使用者作出尚稱滿意的分類，因此本研究設計的註冊流程，可提供使用者結合網路廣告分類機制的參考。

關鍵字

網路廣告 ； 虛擬社群 ； 虛擬寵物 ； 分群機制 ； 支援向量機

並列摘要

Recently, following the rapid developments in network and computer technology, the network based media and users are gradually consolidated in a form called Web2.0. Internet advertising is not only expanding accordingly, but also evolving various new advertising models. One of these models is a new type of Internet advertising mechanism by advertizing on the virtual community. This trend of innovative mechanism also extends to virtual pet based Internet community. However, the behavioral characteristics of virtual pet users in virtual community are rarely effectively utilized in Internet advertising.
This study surveys and reviews some typical cases of current existing virtual pet community platform. To further investigate their background and services, and analyze the properties of registered members, we conduct a preliminary study about these platforms. The classification mechanism of the advertizing category, the users and the facilities for virtual pets are explored. We also build a virtual community platform with the design of an elaborated user’s registration procedure to elicit the users’ interestingness. We also try to establish a matching mechanism of users’ preference and Internet advertising by using SVM(Support Vector Machine) to cluster these users. With the utility of the public Libsvm tool, we are able to do some experiments on the effectiveness of our design. The outcome shows that our design of user’s registration procedure with the automatic clustering of users into three and six clusters could reflect satisfactory classification. This result could be a referential basis for better matching between users and Internet advertising.

並列關鍵字

Internet Advertisement ； Virtual Community ； Virtual Pet ； Clustering Mechanism ； SVM

參考文獻

模式之研究。國立中山大學企業管理學系研究所。

[5]廖英翔、林博文 (2007)。部落格廣告商業模式研究。國立清華大學

the Influential Bloggers in a Community. WSDM’08.

[14]Yifan Chen , Gui-Rong Xue and Yong Yu. (2008) . Advertising Keyword Suggestion Based on Concept Hierarchy.WSDM’08.

[15]Hagel, J. I., and Armstrong, A. M. (1997). Net gain: Expanding makets throgu virtual communities. Boston: HBS Press.