Values with non-word boundaries are not matched by parser

[ybushmanova]

By default, parser wraps text tokens in word boundary tags (\b). Regexp generated this way does not match source text when token starts or ends with non-word character:

source:
hello, my mood is ok, +depression
value: yes, value variant: +depression
Regex: /\b+depression\b/

http://rubular.com/r/8Yc6evs0U3