I’ve set autoCorrectCorpus to: 0.99-1.01-4000-14 and this used to work to bring the norm to 1.0 each night.
Yet the rebuild report says: Mar-30-16 01:17:13 Spam Weight : 197,392 Mar-30-16 01:17:13 Not-Spam Weight: 2,150,448 Mar-30-16 01:17:13 Corpus norm: 0.0918 - (warning: extremely ham heavy) Mar-30-16 01:17:13 Corpus confidence: 0.07542170 Mar-30-16 01:17:13 Corpus norm should be between 0.6 and 1.4 Mar-30-16 01:17:13 Recommendation: You need more spam messages in the corpus. Spam folder only has 361 files in it, but used to have thousands. I haven’t changed anything in the Collecting section, but I’ve been updating to each new build, and am now on 16089. Collected spam has been going up today and is now at 388. Has anything changed code wise that would cause such loss of collected files? Thanks, James. ------------------------------------------------------------------------------ Transform Data into Opportunity. Accelerate data analysis in your applications with Intel Data Analytics Acceleration Library. Click to learn more. http://pubads.g.doubleclick.net/gampad/clk?id=278785471&iu=/4140 _______________________________________________ Assp-test mailing list Assp-test@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/assp-test
