Issues with previously registered vocab terms and regex about adapt HOT 2 CLOSED

This is behavior precisely as expected. Adapt's strength lies in it's abilities as a KET (Known Entity Tagger). Regexes are present to represent information that cannot be known ahead of time (for example, the contents of a text message or web search query). Adapt attempts to compensate for the egregious nature of wildcard regexes (the 3 .*'s) by weighting known entities higher in parse results. The above will always be treated as "lesser" than most parse results containing known entities.

The proper way to model the above would be with a list of known (and probably static) list of ShowVerbs, a static or dynamic list of ShopLocations (is this all lists? can we come up with a better name?), and then a single wildcard regex to capture the items being added. perhaps something like (?P<ShopItem>.*) (to|onto|in), for the english regex. (I think that's valid syntax, don't quote me on it).

TL/DR; Modeling language with regular expressions is asking for disappointment. Use Adapt! It's better ™

from adapt.

Alright, gotcha. Thanks for the clarification.

And yeah, I realized the issues inherent in using that kind of regex, and was going to suggest that he take out at least the shop verb into a list of vocab, if not the location. I just didn't realize that was actually reflected in Adapt.

from adapt.