When I dump my bayes database with sa-learn (--dump data) I find I there are some very large postitive and negative integers in the 2nd and 3rd column which I think are supposed to be the number of ham and spam messages in which the word was found. Here is what I mean
sa-learn --dump data 0.450 -956038913 -368836154 1080161323 anticipate 0.087 134217962 1359217099 1080161028 farmers 0.527 -268107494 -822017837 1080159685 anti 0.945 302055530 16777216 1080158664 strawflower 0.357 2030043308 -771423431 1080161987 emailing 0.119 117442525 838860874 1080160931 profitability 0.088 16777259 167772193 1080161746 448 0.999 56 0 1080161733 sk:a10.tek 0.424 1040516151 1360004724 1080161550 largest 0.000 0 1594492030 1080161636 H*r:66.218.67 etc Does this mean bayes_toks is now corrupt, or is it just something that can be ignored? And if these entries are bad is there some way to dump and clean and import rather than deleting and starting all over again? Any advice would be greatly appreciated. I'm running SA 2.63, perl 5.8.2 and db 4.2.52 BTW. - rick
