1V.Sowmya Sri, K.Punith, Praveen Tumuluru


As E-mail is the fastest mode of sending information all over the world, many people are trying their best to misuse the services. This misuse includes sending of fake emails also known as spam mails. It is either not easy to detect what is spam email and delete the emails because its hard to find out spam emails among them. Many people without having knowledge about what is an spam email when they come across such mails they either delete or remain same in their mail list. Mainly due to marketing and advertising agents emails send by them are irregular and contains unwanted information. These mails cause more storage in mails and time waste. There are some mails which can hack our system by clicking any links provided by them. By this many of the email accounts get hacked and cause great loss to the particular person. E- mail has established a major place in users life. Mails square measure used a significant and vital mode of communication in each personal and skilled aspects of ones life. The fast increase within the variety of account holders over the past number of decades conjointly the increase in mail volume have also made some serious issues. The communications square measure called ham and spam emails. Spam emails square measure is spreading at a huge place from the past few decades. Such spam emails square measure unlawful and unwanted emails that will contain virus, malicious codes, ads or threats. Machine Learning wont allow to screen the spam email mechanically at awfully sensible peace current days.This major problem has created a necessity for reliable and economical anti-spam filters that split the email into spam or ham messages. Spam filters keep the user from delivery spam emails into the inbox. Email spam filters can filter emails either on content or header base. Specific spam filters square measure is classified into 2 teams, particularly machine learning and non-machine learning. In this paper hybrid algorithms are used to detect spam emails. Machine Learning algorithms such as AdaBoost classifier, Gradient Boosting Classifier, Count Vectorizer and Naive Bayes Classifier are some of the algorithms used in prediction of spam emails. . By calculating the accuracy of each and every algorithm on given information the algorithms with high accuracy are combined for further process. A mail id with password is provided and lists of mails are also provided for detecting spam messages. After detecting it provides output with a graph providing the number of ham and number of spam messages detected.


E-mail Spam, Classification, Spam Filter.

Paper Details
IssueIssue 6