How is the score calculated in the term suggester?
I assume it is a combination of term frequency and string distance. However I am seeing some surprising results:
Searching for "chreme" in my index with a term suggester (default settings) suggests "chrome" (score 0.833, frequency 48) and "creme" (score 0.8, frequency 2050). I believe both of these suggestions have an edit distance of 1, so how does the low frequency suggestion get a higher score?