I would like to use Redisearch without using any stemming, because I would like to use the lemmatization of words using external NLP libraries (Nltk, Spacy, …). My application would load Redis with the data already lemmatized, while for the searches with redisearch, the same application would undergo the same process of lemmatization of the query string.
All this because of the quality offered by the stemming algorithm used by Redisearch, the Snowball, which with the Italian and French languages can only have an academic value, not at all acceptable for a solution of practical use because of the excessive false positives it can generate. Is there a way to completely disable the stemming to avoid that the default English language stemming is still used?