# assuming the raw data is in $line.
$line =~ m/^[^;]*;\s*([^;]*?)\s*;/

# $1 now holds whatever is between the second and third 
# semicolon, leading and trailing spaces trimmed.