以統計方法進行Facebook盜用行為偵測

Abstract

社群網站上的個人資料是重要的課題，因為一旦社群網站的個人帳號被盜用，所有在上面的個人資料都會被第三者取得，不論帳號擁有者做過任何隱私權設定。因此，本篇論文以統計方法並使用Support Vector Machine (SVM)，進行臉書的盜用行為偵測。經由分析使用者在線上的瀏覽紀錄，可以發現正常的使用者在社群網站的行為比較主動，盜用帳號者偏好閱讀私人訊息。

Privacy of personal information on social networking websites has become an important issue, because when a social networking website account is used by a person other than the owner, all personal data stored on the website can be retrieved, no matter how the owner sets the privacy options. Therefore, this paper proposes a statistical approach with the use of Support Vector Machine (SVM) to detect whether the Facebook account user is the actual owner. By analyzing online browsing behavior features, it is found that the normal user tends to be more active and that the stealthy user prefers to read personal messages.