如何在Elasticsearch中仅返回Should查询的最高分记录

Got it, let's tackle this problem. You want to return only the highest-scoring documents from your bool/should query without using any must clauses—here are a couple of solid approaches:

Method 1: Two-Step Query (Simple & Direct) #

First, run your original query to grab the max_score value (you can set size: 0 to skip returning hits, since we only need the max score):

GET testindex1/_search
{
  "query": {
    "bool": {
      "should": [
        { "query_string": { "default_field": "name", "query": "xyz", "default_operator": "AND" } },
        { "query_string": { "default_field": "description", "query": "*", "default_operator": "AND" } },
        { "query_string": { "default_field": "place", "query": "*", "default_operator": "AND" } }
      ]
    }
  },
  "size": 0
}

This will return the max_score (in your case, 2.287682) in the response.

Then, run a second query using a post_filter with a script to only keep documents where the score matches the max score:

GET testindex1/_search
{
  "query": {
    "bool": {
      "should": [
        { "query_string": { "default_field": "name", "query": "xyz", "default_operator": "AND" } },
        { "query_string": { "default_field": "description", "query": "*", "default_operator": "AND" } },
        { "query_string": { "default_field": "place", "query": "*", "default_operator": "AND" } }
      ]
    }
  },
  "post_filter": {
    "script": {
      "source": "_score == params.max_score",
      "params": {
        "max_score": 2.287682
      }
    }
  }
}

This will return only the documents with the highest score, and we didn't use any must clauses—perfect for your requirement.

Method 2: Single Query with Aggregations (App Layer Filtering) #

If you prefer a single round trip to Elasticsearch, use a top_hits aggregation to fetch all matching documents sorted by score, then filter for the highest-scoring entries in your application code:

GET testindex1/_search
{
  "query": {
    "bool": {
      "should": [
        { "query_string": { "default_field": "name", "query": "xyz", "default_operator": "AND" } },
        { "query_string": { "default_field": "description", "query": "*", "default_operator": "AND" } },
        { "query_string": { "default_field": "place", "query": "*", "default_operator": "AND" } }
      ]
    }
  },
  "aggs": {
    "top_matching_docs": {
      "top_hits": {
        "sort": [{"_score": "desc"}],
        "size": 10000 // Set this to a value larger than your total matching docs
      }
    }
  },
  "size": 0 // Skip the default hits array, we'll use the aggregation results
}

In your app, take the first document's _score from aggregations.top_matching_docs.hits.hits as the maximum score, then filter the rest of the hits to keep only those with that score.

Notes #

• For Method 1, make sure dynamic scripting is enabled in Elasticsearch (it's allowed by default unless you've modified security settings).
• For Method 2, set size in the top_hits aggregation to a number that's guaranteed to cover all matching documents in your index to avoid missing any high-scoring entries.

内容的提问来源于stack exchange，提问作者G2K