Evaluation of Time Complexities of Bayesian Vs Hybridized Word Stemming Techniques for Advanced Fee Fraud Emails Filtering

Abstract

Okunade Oluwasogo,  Adekunle,  Afolorunso Adenrele and Adebayo Adegboyega

Time execution of content-based spam filter was investigated using the Bayesian statistical algorithm against Bayesian statistical algorithm incorporated with a word stemming. The execution time intervals for the algorithms implementation of the two techniques were evaluated by subjecting the filters to manipulated and non-manipulate spam mails. The experiment shown that both single technique (Bayesian) and combined techniques (Bayesian incorporated with word stemming) executed suspicious terms manipulated mails faster (within a short time) compared to non-manipulate suspicious terms mails. Combined algorithms performed better and faster in a sophisticated and manipulated environment. The algorithm is more rugged and performed better when suspicious term/tokens were manipulated to deceit the filter.