Justin Mason <[EMAIL PROTECTED]> writes: >> remove Flex Hex rules due to low accuracy
> what were the results? http://www.pathname.com/~quinlan/DETAILS.new 0.410 0.3699 1.2903 0.223 0.45 0.01 T_HTML_COLOR_FLEX_HEX_3 0.466 0.4103 1.6762 0.197 0.41 0.01 T_HTML_COLOR_FLEX_HEX_1 0.466 0.4103 1.6762 0.197 0.41 0.01 T_HTML_COLOR_FLEX_HEX_2 and if I look at per-user results on HTML messages: 0.466 0.4103 1.6762 0.197 0.41 0.01 T_HTML_COLOR_FLEX_HEX_1 0.502 0.2352 1.2411 0.159 0.36 0.01 T_HTML_COLOR_FLEX_HEX_1:cthielen 1.661 1.6682 0.0000 1.000 0.70 0.01 T_HTML_COLOR_FLEX_HEX_1:daf 0.038 0.0395 0.0000 1.000 0.42 0.01 T_HTML_COLOR_FLEX_HEX_1:jm 0.526 0.5422 0.0000 1.000 0.62 0.01 T_HTML_COLOR_FLEX_HEX_1:quinlan 0.182 0.1009 0.9072 0.100 0.20 0.01 T_HTML_COLOR_FLEX_HEX_1:rODbegbie 0.197 0.0964 2.6475 0.035 0.16 0.01 T_HTML_COLOR_FLEX_HEX_1:theo I added leading/trailing whitespace trimming to the code today (found during some more testing) and it's possible that would fix some of the ham hits, but I really doubt it's worth more. It might be better to go after specific patterns, but it doesn't really seem like a very fertile area. #FaFtFw #FaFuFb #FgFhFu #FlFlFx #FoFcFf #FoFnFs #FrFnFb #FwFjFz #FxFeFu #fffffg Daniel -- Daniel Quinlan http://www.pathname.com/~quinlan/
