WORKLOAD CHARACTERIZATION OF SPAM EMAIL FILTERING SYSTEMS
Yan Luo
Department of Electrical and Computer Engineering, University of Massachusetts Lowell, Massachusetts, USA
ABSTRACT
Email systems have suffered from degraded quality of service due to rampant spam, phishing and fraudulent emails. This is partly because the classification speed of email filtering systems falls far behind the requirements of email service providers. We are motivated to address this issue from the perspective of computer architecture support. In this paper, as the first step towards novel architecture designs, we present extensive performance data collected from measurement and profiling experiments using representative email filtering systems including CRM114, DSPAM, SpamAssassin and TREC Bogofilter. We provide detailed analysis of the time consuming functions in the systems under study. We also show how the processor architecture parameters affect the performance of these email filters through simulation experiments.
KEYWORDS
Workload Characterization, Email Filtering, Anti-Spam
Original Source Link : http://airccse.org/journal/nsa/1010s3.pdf
No comments:
Post a Comment