Using Cellular Automata for Improving KNN Based Spam Filtering

As rapid growth over the Internet nowadays, electronic mail (e-mails) has become a popular communication tool. However, junk mail also, known as spam has increasingly become a part of life for users as well as internet service providers. To address this problem, many solutions have been proposed in the last decade. Currently, content-based anti-spam filtering methods are an important issue; the spam filtering is considered as a special case of binary text categorization. Many machine learning techniques have been developed and applied to classify email as spam or non-spam. In this paper, we proposed an enhanced K-Nearest Neighbours (KNN) method called Cellular Automaton Combined with KNN (CA-KNN) for spam filtering. In our proposed method, a cellular automaton is used to identify which instances in training set should be selected to classify a new e-mail; CA-KNN selects the nearest neighbours not from the whole training set, but only from a reduced subset selected by a cellular automaton.