智慧型電子郵件管理系統

Abstract

近來由於網路的發達，使得每個人所要面對的資訊量也越來越多，一些資訊管理系統也被用來減輕人們的負擔。但是大部分的系統必須將資訊由網路上下傳後才能管理，這樣會造成網路頻寬的浪費，因為即使是使用者不想要的資訊也一併的被傳送。 在本篇論文中，我們開發了一套智慧型電子郵件管理系統（IEMS）用來幫助使用者管理電子郵件。IEMS是一套包含伺服器端、客戶端以及規則庫的電子郵件管理系統。透過資料探勘的機制，系統能夠由使用者的使用習慣中自動找出管理規則。在客戶端取回信件之前，系統會對每一封信件做出一個推論，這樣在客戶端取回信件時，即可決定是否將信件過濾，以減輕網路的負擔。使用者對這些推論的採用與否，能提供回饋來調整系統。系統中結合了規則式推論以及案例式推論兩種方法，以提高預測正確率。另外使用者的管理規則是由規則庫集中管理，專家可以經由規則庫提供使用者一些信件管理上的知識，幫助使用者管理電子郵件。實驗結果顯示IEMS可以達到約百分之八十的推導正確率，能夠正確的幫助使用者管理信件而減輕使用者負擔。

As network becomes more and more popular today, the information people can get from the Internet is increasing rapidly. Some previous systems developed to help people manage their information have to download the information first and then can manage it. This would cause the waste of network traffic because of the transmission of the undesired information. In this thesis, we develop an e-mail management system named Intelligent E-mail Management System (IEMS) including the e-mail server, client and a Rule Base to help users manage their e-mails. A mining algorithm is used in the system to find managing rules from user’s behavior automatically. The prediction of a mail is made before transferring it to client; therefore, the undesired mails can be filtered at server to save network traffic. User’s decisions on the predictions made by the system will provide feedback to refine the system. The rule-based and instance-based inference processes are combined to increase the accuracy of predictions. Moreover, rules are centralized maintained by the central Rule Base; thus, new knowledge can be provided by experts to help users manage their mails. Experimental results show that IEMS can infer E-mail reading preference of individual user correctly up to the accuracy about 80%. We may conclude that IEMS can help users manage their mail messages effectively.