Question

Do you know any big enough lemmatizer database that returns correct result for following sample words:

geese: goose
plantes: //not found

Wordnet’s morphological analyzer is not sufficient, since it gives the following incorrect results:

geese: //not found
plantes: plant

Answer 1

MorphAdorner seems to be better at this, but it still finds the incorrect result for “plantes”

plantes: plante
geese: goose

Maybe you’d like to use MorphAdorner to do the lemmatization, and then check its results against WordNet. You can use the WordNet API to perform lookups without first performing lemmatization by calling findtheinfo_ds. This allows you to use a lemmatizer like MorphAdorner first. (If you wanted to use the lemmaitzer, you’d need to call morph separately and call findtheinfo_ds on the lemmas that it returned.)

On the other hand, I only spent about 5 seconds looking at MorphAdorner for this purpose, and there may be a way to eliminate the incorrect “plantes” answer without having to use any other outside resource.