elasticsearch：使用ngram分析器时避免重复计分

Question

Suppose I search for "hello" when the document contains "hello" and "hello hello" I want "hello" to have higher scoring.

I am using ngram index and search analyzer. (Because I really need this for other scenarios) So "hello hello" gets matched twice and hence shows as the top result. Is there any way I can avoid this? I have already tried term query, match phrase query, multi match queries all of them scores "hello hello" higher.

Answer 1

I solved this by adding a duplicate unanalyzed (keyword) column for the document and used bool clause to boost the term query.

var res = client.Search<MyClass>(s => s
  .Query(q => q
    .Bool(
        b1 => b1.Should(
            s1 =>s1
            .Term(m=>m
                .Field(f => f._DUPLICATE_COLUMN)
                .Value("hello")
                .Boost(1)
            ),

            s1=>s1.Match(m => m
            .Field(f => f.MY_COLUMN)
            .Query("hello")
            .Analyzer("myNgramSearchAnalyzer")
            )
        )
        .MinimumShouldMatch(1)
    )
  )
);