Hello, We are trying to use NGramFilterFactory for approximative search with solr 7.
We usually use a similarity with no tf, no idf (our similarity extends ClassicSimilarity, with tf and idf functions always returning 1). For ngram search though, it seems inappropriate since it scores a word matching with one ngram the same as a word matching with, let's say, seven ngrams. We would like a similarity that gives a higher score to a document matching more ngrams, but not using term frequency (we have multivalued fields, and a word might be repeated in more than one entry of our multivalued field, but we dont want that document to get a higher score because of that) Does anyone have experienced the same issues? Best regards, Elisabeth