[Talk-GB] OS Locator / OSM correspondence list generation

Robert Scott lists at humanleg.org.uk
Thu May 13 17:42:10 BST 2010


On Thursday 13 May 2010, Emilie Laffray wrote:
> What about using double metaphone for finding spelling disagreements?
> 
> Emilie Laffray

It's something I looked at briefly and depending on how many error reports I get I may look at it again if I find lots of phonetic-type errors that aren't matched properly. Postgres' fuzzystrmatch module has DM functions too which is useful.

DM as far as I can see is designed for single words, and this would then multiply the word splitting problem I already have.

If there were another mechanism I would like to introduce it would be longest common substring matching to catch situations where words have been reordered.


robert.





More information about the Talk-GB mailing list