Multiple inconsistent street type faults about ols-geocoder HOT 5 OPEN

This one is particularly weird because the street type was spelled wrong, and then not matched, and then matched as a suffix instead of a prefix... basically, the "st" of street is becoming the "st" of saint, and that is enough of a partial match that the word king is ignored... I think we may need some special handling for "st", but we also need to change something in the scoring or parsing because that combination of faults isn't sensible.

from ols-geocoder.

estimate is to fix this specific case, more general "meta-analysis of faults" is the sort of thing we would time-box to how much you want to spend as there is no shortage of faults to analyze.

from ols-geocoder.

Please add appropriate estimate label (e.g., estimate 2 hours) for the specific fix.

from ols-geocoder.

I set the estimate to 21 for specifically fixing the handling of multiple inconsistent street-type faults. I think I may need to change the order of some logic. Much testing is required.

Maybe that is what you meant by "meta-analysis of faults" ... I took it to mean a much broader analysis of which this is was just one example of something weird. When we do a broader analysis type of thing, we usually time box it, say, lets spend 2 weeks reviewing these errors and try to find the easiest/ most effective fixes/tweaks/etc.

from ols-geocoder.

Perhaps all we really need to do is add a special case for not prefix-matching "st" (as in the street type) to "st" as in saint patrick's, because "st" is way too common to be used in that way. This actually seems to be a pretty pervasive problem.

from ols-geocoder.