Precise definition depends on the language.

Mastering Regular Expressions, 2nd Ed., Jeffery E. F. Friedll, published by O'Reilly characterizes\w in its "Common Metacharacters..." chapter, this way:

Part-of_word character   Often the same as [a-zA-Z0-9_], although some ools omit the underscore, while others include all the extraalphanumerics characters in the locale. If Unicode is supported, \w usually refers to all alphanumerics (notable exception: Sun's Java regex package whose \w is exactly [a-zA-Z0-9_</c>).

• \p{IsWord} for Perl
• and as [A-Za-z0-9_] for Java.

Regretably, the definition of \p{isWord} -- [_\p{L1}\p{Lu}\p{Lt}\p{Lo}\p{Nd} -- is, for me, almost impenetrable but Friedll's characterization may be as good as you'll get without deep study of perlretut and friends.