ES gives very different scores, in match_phrase_prefix, for similar documents even I use DfsQueryThenFetch

Hi everyone,

I have some problems about phrase prefix search on ES.

I have an index which has these documents:

id:1 -> ABC DEF XYZ QWE

id:2 -> XYZ QWE ABC DEF

id:3 -> ABC XYZ STH XYZ QWE

id:4 -> KLM PRE TUV JKL

id:5 -> KLM JKL PRE TUV …(40-Million records in total)

when I invoke a search with query like this:

{
{

"bool" : {

"must" : {

 "match" : {

   "fullString" : {

     "query" : "ABC DEF XYZ QWE",

     "type" : "boolean",

     "operator" : "OR",

     "cutoff_frequency" : 0.01

   }

 }

},

"should" : [ {

 "match" : {

   "fullString" : {

     "query" : "ABC DEF XYZ QWE",

     "type" : "phrase"

   }

 }

}, {

 "match" : {

   "fullString" : {

     "query" : "ABC DEF XYZ QWE",

     "type" : "phrase_prefix",

     "max_expansions" : 3

   }

 }

} ]

}

}

id:1 -> ABC DEF XYZ QWE -> score: 58.6

id:2 -> XYZ QWE ABC DEF -> score: 0.56

id:3 -> ABC XYZ STH XYZ QWE -> score: 0.53

and when I query = "PRE TUV"

id:4 -> KLM PRE TUV JKL

id:5 -> KLM JKL PRE TUV these documents have very unlikely scores. Is
there anyone who could explain me or give a link of detailed documentation
about this problem?

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/elasticsearch/3b814721-0861-41c9-9db0-e2310be009cd%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.