Patrick Gaskin (@pgaskin)
1d05618 Improve word normalization for LookupWord[s]
In order (checked after each one):
* Trim leading/trailing spaces.
* Collapse all whitespace into a single space.
* Trim leading/trailing opening/closing unicode punctuation.
* Replace all unicode dash-like characters with a dash.
* Collapse multiple dashes into a single one.
* Try each of the following (without chaining):
* Attempt to stem the word.
* Try removing just -s*.
* Try removing just -ly, then -ing.
* Try again, but fold all unicode characters into their
normalized form.