word counting code does not account for & being special html symbol. about boilerpipe HOT 2 CLOSED

adding '&' to the PAT_NOT_WORD_BOUNDARY of UnicodeTokenizer gives the better 
output. Of course then it is not really a UnicodeTokenizer but a HtmlTokenizer. 
It might be better to combine the regexp of that tokenizer class and the isWord 
method into a new class HtmlWordCounter giving it a public static method which 
does the word counting so that other projects can easily reuse it. 

The input to UnicodeTokenizer is Unicode text, not HTML-escaped text. If you 
want to use UnicodeTokenizer, you have to prepare the input appropriately.

As you have pointed out, you want a HtmlTokenizer. Boilerpipe takes care of 
HTML entity resolution via SAX parsing, so there is no need to replicate that 
functionality here.

Marking as WontFix.