Identifying junk electronic mail in Microsoft outlook with a support vector machine

Abstract

Abstract

Email Spam Detection Using Combination of Particle Swarm Optimization and Artificial Neural Network and Support Vector Machine

“…The BDT learning time is too long. Woitaszek and colleagues [9] of a SVM simple and a personal dictionary to determine the e-mails are commercially used.…”

Using GMDH-based networks for improved spam detection and email feature analysis

“…The number of selected attributes is 10, 10, and 9, respectively. Eight attributes {7, 16,21,23,25,46,52, and 53} are common to all three input subsets. This indicates the effectiveness of such attributes as spam predictors and the robustness of the modeling process.…”

“…Table 6 shows the structure of the resulting models for the 3 ensemble members together with the classification accuracy for each individual model on both the training set (948 cases) and evaluation set (1757 cases). The models select between 9 and 11 attributes as inputs, with 7 attributes {7, 16,25,27,46, 52, and 53} being common to all three models. Out of this subset of 7 attributes, six attributes {7, 16, 25, 46, 52, and 53} were also common to the models in Table 3.…”

“…Therefore, they can be more effective in dealing with spammers' tactics. Recently, various machine-learning techniques have been used for spam filtering, including support vector machines [13,22,46], memory-based learning [6,35,41], Ripper rule-based learning [12,33], boosting decision trees [10], rough sets [49], neural networks [11,48], Bayesian classifiers [5,6,23,33,34,36,37], and fuzzy logic [17,32]. Among these methods, Bayesian classifiers have been widely applied as one of the most effective spam filters [19,23].…”

An Empirical Performance Comparison of Machine Learning Methods for Spam E-Mail Categorization

“…According the description given in Woitaszek et al [11], an e-mail may be represented by a feature vector x that is composed of the various words from a dictionary formed by analyzing the collected e-mails. Thus, an e-mail is classified as spam or non-spam by performing a simple dot product between the features of an e-mail and the SVM model weight vector,…”