As the man said, we need your data to be able to tell something about this. From looking at your regex (mangled by the fact that it's not in code tags), it seems like it is not looking for a word, but for a word preceded and followed by a % or a digit. Is that what you intend to do?
On PerlMonks, put <c>..</c> tags around your code. It's like <pre>..</pre>, but it handles escaping special characters for your (such as &, <, >, [ and ]). It also wraps long lines while offering a link to download the unwrapped version.