A TREC along the Spam Track with SpamBayes
Tony Meyer · Text REtrieval Conference · 2005
This paper describes the SpamBayes submissions made to the Spam Track of the 2005 Text Retrieval Conference (TREC). SpamBayes is briefly introduced, but the paper focuses more on how the submissions differ from the standard installation. Unlike in the majority of earlier publications evaluating the effectiveness of SpamBayes, the fundamental ‘unsure’ range is discussed, and the method of removing the range is outlined. Finally, an analysis of the results of the running the four submissions through the Spam Track ‘jig’ with the three private corpora and one public corpus is made. 1 SpamBayes SpamBayes [1] was born on August 19th 2002, soon after publication of A Plan for Spam [2]; Tim Peters and others involved with the Python development community developed code based on Graham’s ideas, with the initial aim of filtering python.org mailing-list traffic, although this quickly progressed to also filtering personal email streams (today, although most python.org mailing lists do use SpamBayes, the overwhelmingly most common use of SpamBayes is for personal email filtering). Although the project initially started with Graham’s original combining scheme, it currently uses the chi-squared combining scheme developed by Robinson.