M.A. Thesis:

Distinguish spam from non-spam emails using data mining

Recently, with the development of new technologies and fast communication tools such as the Internet ,e-mail it is considered as one of the most important tools for sending and receiving information. In this regard, the cheapness of this technology has made the bed It is suitable for people who are profitable for various purposes, such as promoting products, promoting opinions, sending various types of malware and viruses to destroy users' systems or penetrating their confidential information, etc. Also, due to the rapid growth of social networks and the uncertainty of identity of the senders of emails, users are widely involved in a series of spam or spam emails that cause waste of network bandwidth and spend time and money for For users, it's very necessary to keep the computers safe because of attacks by these spam. The traditional signature-based signature-based spam detection method is not effective due to new user tricks. Today, spam detection and mail ordering problems are solved using a machine learning and data mining system. In the method described, we need to detect spam from non-spam and then sort the emails to reduce the error rate, and then filter spam at various levels can be applied. Based on a series of features, selecting the appropriate features will greatly affect the functionality of the category. Considering that the reduction of the attribute can be in two ways: extracting the attribute and selecting the attribute, based on a specific criterion, in this category, along with familiarization with different methods of detection and prevention of spam, in a proper way for categorizing spam Using modern data-driven techniques, we present a way to extract the proper features to reduce the volume of data and increase the speed of the operation of spam categories.

Keywords: specific criterion Data mining, Spam. Classification,