Comparative study on email filtering algorithms
Taifeng Liu · Jisuanji gongcheng yu sheji · 2008
The architecture of email filters is given,and six algorithms for blocking spam is introduced.On four public available datasets,we do a full comparative study on the six algorithms by using three evaluation criteria.The main conclusions drawn from the experimental results are firstly that combing multiple filters can gain better performance than a single filter.Secondly,comparing with other two datasets,on PU1 and Ling-spam,a filter shows more optimistical results.The performance of NB and k-NN is greatly decided by datasets.The paper is significative for a researcher who wants to study email filtering,improve or design an email algorithm.