README.Rmd

output: github_document

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.path = "man/figures/README-",
  out.width = "100%"
)
library(topomatch)

topomatch

Helper function for matching toponyms from different sources, that can be written in slightly different ways. Allows to inspect the matching and act accordingly.

countries1 <- spData::world$name_long
countries2 <- unique(maps::world.cities$country.etc)

(country_matches <- topomatch(countries1, countries2))

There are some manual fixes needed for those toponyms that weren't correctly matched. Just write the fixes in a named vector. If there is no correct match for one toponym, give it an NA.

## Inspect the competing candidates for the unmatched countries
(bm <- best_matches(country_matches)[unmatched(country_matches)])

cnames_fixes <- setNames(
  c("Congo Democratic Republic", NA, "Laos", "Korea North",
    "Korea South", NA),
  names(bm)
)

## Fix the incorrectly matches from similarity as well
cnames_fixes <- c(
  cnames_fixes,
  "United States" = "USA",
  "French Southern and Antarctic Lands" = "France",
  "Côte d'Ivoire" = "Ivory Coast",
  "United Kingdom" = "UK",
  "Antarctica" = NA,
  "Northern Cyprus" = "Cyprus",
  "Somaliland" = "Somalia",
  "South Sudan" = "Sudan"
)

Now you can transcribe the original toponyms to the matched terms.

translate <- transcribe(country_matches, fixes = cnames_fixes)

translate(c("United Kingdom", "Kosovo"))

## "Translate" all of the original toponyms
countries1_trans <- translate(countries1)

## Only those "fixed" as NA are not found in the second list
countries1[!countries1_trans %in% countries2]

Method

Wraps local-global alignment algorithm borrwed from bioConductor package Biostrings. Works better than global alignment and requires less fine-tuning (although is considerably slower too) https://ro-che.info/articles/2016-12-11-local-alignment.

Installation

remotes::install_gitlab("umr-astre/topomatch", host = "forgemia.inra.fr")