I don't know the exact term, I would like to see more suggestions from suggester. Here some data:
"ipad not expensive" "ipad red" "buy ipad"
So here "red" is a weak word, whereas "ipad" is strong.
And results I would like:
"ipad red" => "ipad red" / "ipad red not expensive" / "buy ipad red"
"red" => nothing
"buy i" => "buy ipad" "buy ipad red"
Here the key is to tokenize query, call several suggester (term1 / term2 / term1+term2 ....) like n-gram, and mix results. I achived this with code, but I am wondering if there is a pure elasticsearch way?