统计词性语料库

open(In,"corpus.txt");
open(out,">cibiao.txt");
while(<In>){
	chomp;
	@Words=$_=~/(\S+)\/\S+/g;
	@POSs=$_=~/\S+\/(\S+)/g;
	for($i=0;$i<@Words;$i++){
		# if(defined $hash{$Word[$i]}){
		# $refPOS=$hash{$Word[$i]};
		# ${$refPOS}{$POSs[$i]}=1;
  # } 
	# else{
		# my %posHash=($POSs[$i],1);
		# $hash{$Words[$i]}=\%posHash;
  # }
	${$hash{$Words[$i]}}{$POSs[$i]}=1;
	}
}
close(In);
foreach $word(sort keys %hash){
	print "$word\t";
	print out "$word\t";
	$pPOS=$hash{$word};
	foreach $pos(sort keys %{$pPOS}){
    print "$pos;";
	print out "$pos;";
   }
	print "\n";
	print out "\n";
}
close(out);