svn commit: rev 10701 - incubator/spamassassin/trunk/masses/tenpass

[qu...@apache.org]

Author: quinlan
Date: Sun May 16 13:25:17 2004
New Revision: 10701

Modified:
   incubator/spamassassin/trunk/masses/tenpass/10pass-compute-tcr
   incubator/spamassassin/trunk/masses/tenpass/10pass-run
   incubator/spamassassin/trunk/masses/tenpass/compute-current-tcr
Log:
get most of the tenpass stuff to the point where it runs


Modified: incubator/spamassassin/trunk/masses/tenpass/10pass-compute-tcr
==============================================================================
--- incubator/spamassassin/trunk/masses/tenpass/10pass-compute-tcr	(original)
+++ incubator/spamassassin/trunk/masses/tenpass/10pass-compute-tcr	Sun May 16 13:25:17 2004
@@ -1,15 +1,17 @@
 #!/bin/sh
 
+SCORESET="0"
+
 for run in 1 2 3 4 5 6 7 8 9 10 
 do
   mkdir tmp/10passrules > /dev/null 2>&1
   cp ../rules/[0-9]*.cf tmp/10passrules
-  ./rewrite-cf-with-new-scores ../rules/50_scores.cf \
+  ./rewrite-cf-with-new-scores $SCORESET ../rules/50_scores.cf \
 	tenpass_results/scores.$run > tmp/10passrules/50_scores.cf
 
   ./fp-fn-statistics --cffile=tmp/10passrules \
 	--spam=tenpass_results/spam.log.$run \
-	--nonspam=tenpass_results/nonspam.log.$run > tmp/stats
+	--nonspam=tenpass_results/ham.log.$run > tmp/stats
 
   grep TCR: tmp/stats
 done

Modified: incubator/spamassassin/trunk/masses/tenpass/10pass-run
==============================================================================
--- incubator/spamassassin/trunk/masses/tenpass/10pass-run	(original)
+++ incubator/spamassassin/trunk/masses/tenpass/10pass-run	Sun May 16 13:25:17 2004
@@ -1,11 +1,13 @@
 #!/bin/sh
 
 # change these!
-NSBASE=../../logs/nonspam-jm
-SPBASE=../../logs/spam-jm
+NSBASE=ham-logs
+SPBASE=spam-logs
+SCORESET="0"
 
 passes="1 2 3 4 5 6 7 8 9 10"
-mkdir tenpass_results
+mkdir -p tenpass_results
+mkdir -p ORIG
 
 > make.output
 
@@ -15,28 +17,28 @@
   echo "Training for corpus $id..."
   pwd; date
 
-  > nonspam.log
-  > spam.log
+  > ORIG/ham-set$SCORESET.log
+  > ORIG/spam-set$SCORESET.log
+
   echo -n "(using corpora blocks: "
   for notid in $passes ; do
     if [ "$notid" != "$id" ] ; then
       echo -n "$notid "
-      cat $NSBASE/split-$notid.log >> nonspam.log
-      cat $SPBASE/split-$notid.log >> spam.log
+      cat $NSBASE/split-$notid.log >> ORIG/ham-set$SCORESET.log
+      cat $SPBASE/split-$notid.log >> ORIG/spam-set$SCORESET.log
     fi
   done
   echo "for training)"
 
   make clean >> make.output
   make >> make.output 2>&1
-  ./evolve
-  pwd; date
+  ./runGA
+  pwd
+  date
 
   echo "Saving test data for corpus $id..."
 
-  cp $NSBASE/split-$id.log tenpass_results/nonspam.log.$id
+  cp $NSBASE/split-$id.log tenpass_results/ham.log.$id
   cp $SPBASE/split-$id.log tenpass_results/spam.log.$id
-  cp craig-evolve.scores tenpass_results/scores.$id
+  cp gen-set$SCORESET.scores tenpass_results/scores.$id
 done
-
-

Modified: incubator/spamassassin/trunk/masses/tenpass/compute-current-tcr
==============================================================================
--- incubator/spamassassin/trunk/masses/tenpass/compute-current-tcr	(original)
+++ incubator/spamassassin/trunk/masses/tenpass/compute-current-tcr	Sun May 16 13:25:17 2004
@@ -9,7 +9,7 @@
 my $sumsp = 0;
 
 open(SPAM, "<spam.log");
-open(NONSPAM, "<nonspam.log");
+open(HAM, "<ham.log");
 open(SCORES, "<../rules/50_scores.cf");
 
 while(<SCORES>)
@@ -41,7 +41,7 @@
 }
 close(SPAM);
 
-while(<NONSPAM>)
+while(<HAM>)
 {
     next if /^\#/;
     /.\s+[-0-9]*\s+[^\s]+\s+([^\s]*)\s*$/;