search

thinkingmachines / linksight-2018

LinkSight is a web app for applying the Philippine Standard Geographic Code to messy and misspelled barangay, municipality, city, and province names.
https://linksight.thinkingmachin.es
GNU General Public License v3.0
11 stars 1 forks source link

Feature/add match type #164

Closed stuckoverflo closed 6 years ago

stuckoverflo commented 6 years ago

Opening for early feedback but don't merge yet.

See #141

marksteve commented 6 years ago

@florobarotjr

screen shot 2018-08-28 at 1 06 17 pm
stuckoverflo commented 6 years ago

Odd, I'm getting exact matche on Sabang on my local copy of the branch :| @marksteve

image

marksteve commented 6 years ago

@florobarotjr I just tried again both in this branch and master. Master gets only 4 partial matches. This branch gets 12.

FamilyDOC Branches - Sheet1.csv.zip

stuckoverflo commented 6 years ago

@marksteve ah, thanks for providing the CSV. I was using a more cleaned up version of that dataset.

Seems like the change is working as intended in that if we can't find a lower interlevel match, we'll provide whatever higher interlevel match we find and 'No match' as an option. Before, those are marked as exact matches even if we weren't able to find the barangay data.

Having said that, the reason why Sabang is not being matched in the example is because of the substring matching :(. The City field is Dasmarinas. Cavite. It found Cavite first in the reference file so it didn't match to Dasmarinas at all.

marksteve commented 6 years ago

Discussed with @florobarotjr that this is an edge case that we shouldn't be handling for now.