Framework for Phishing Detection in Email under Heave Using Conceptual Similarity

Abstract

Today everything is available in online. Every day so many users start their online transactions. The main reason behind thisis number of alternatives and best deals are available there. They can choose according to their taste with cost effective manner. This is one side of a coin. The other side fully dealt with security problems and frauds in the online transactions. Among most of the online transactions email is the shortcut and flexible for both communication as well as for attack. So this paper mainly focuses on detection of phishing attacks and categorizes the emails based on specified and critical properties which give more information about the source of the phishing. In general most of the existing systems focus on email classification based on header part or body part. Most of the filters available today focus mainly on mail headers only. Sometimes this is not enough to detect the fraud. Some more studies focus on body part also. But they follow document clustering with term intensive similarity. First, to identify advanced phishing attacks blind term intensive similarity is not sufficient. Second, emails system is like online stream. So the nature of the phishing behavior may change time to time. In that case online learning is also required to handle concept drifts. This paper focuses on conceptual similarity along with term intensive similarity. We introduced a novel procedure named as “Framework for Phishing detection in email under heave using conceptual similarity†to adaptively classify the emails. Simulation results shows that our proposed approach effectively detect and isolate the emails with phishing attack by comparing underlying concept.