I have a Rails application with a Postgres backend.
I need to add a full-text search that will allow fuzzy searches based on Levenshtein distance or other similar metrics. Add the fact that lexer / stemmer should work with non-English words (it would be nice to just turn off language-dependent functions when lexing, so as not to be confused with the target language, which may have significant words that are considered irrelevant by the English engine).
I think tsearch Postgres will not be applied here because it does not have a fuzzy search - please correct me if I am wrong.
What are the possible combinations of backends and plugins? He would prefer solutions that add less infrastructure (for example, if Postgres might have fuzzy fts, why use external Lucene); OTOH, the inclusion of Rails plugins is important.
What would you recommend?
update : it looks like I need metrics based on n-grams rather than Levenshtein.
source
share