Make Sorting Office smarter with spelling

[giacecco]

Today Sorting Office will recognise "St. Ives" as a town, but not "St Ives", without the dot. I believe we can be smarter than that! :-)

I suggest we do the following:

• work on defining of a series of spelling transformations on top of the ones we do already, e.g. we're case insensitive at the moment, but we may also ignore dots, hyphens and punctuation in general when searching for matches in the reference tables, and then
• implement the above in Sorting Office

@MurrayData can help not falling into any traps in doing this. @pezholio what do you reckon?

See also OpenAddressesUK/forum#2 .

[pezholio]

We handle this with streets already, this would probably need some Elasticsearch configuration to ignore punctuation

[giacecco]

I believe we need to see the problem independently from the implementation. There are many other cases we need to manage like abbreviations in general, e.g. see OpenAddressesUK/forum#2 . If we find out that Elastic Search is not good enough, we'll have to think to something else.