Master Elasticsearch Queries: Boost, Phrase, Multi‑Match & Query String Explained

Match Query

Use a bool query with should clauses to combine several match queries, assigning a boost value to each field so that matches in higher‑boosted fields contribute more to the relevance score.

{
  "query": {
    "bool": {
      "should": [
        {
          "match": {
            "product_name": {
              "query": "apple",
              "boost": 3
            }
          }
        },
        {
          "match": {
            "description": {
              "query": "apple",
              "boost": 1
            }
          }
        }
      ]
    }
  }
}

The query searches the product_name and description fields for the term "apple"; the product_name match is weighted three times higher than the description match.

Exact Phrase Match (match_phrase)

The match_phrase query forces the terms to appear in the specified order. A simple PHP client example searches the title field for an exact phrase.

$client = ElasticSearchCLient::getInstance();
$indexParams = [
    'index' => 'resty_product_test_index',
    'body' => [
        'query' => [
            'match_phrase' => [
                'title' => '开源技术小栈20245直播间'
            ]
        ]
    ]
];
$indexResponse = $client->search($indexParams);

A sample hit returned by this query:

"hits": [
    {
        "_index": "resty_product_test_index",
        "_type": "_doc",
        "_id": "20245",
        "_score": 2.3340414,
        "_source": {
            "id": 20245,
            "title": "开源技术小栈20245直播间",
            "content": "开源技术小栈-腾讯云开发者社区，开源技术小栈20245直播间"
        }
    }
]

Adding a slop parameter relaxes the positional constraint, allowing a limited number of intervening terms.

$client = ElasticSearchCLient::getInstance();
$indexParams = [
    'index' => 'resty_product_test_index',
    'body' => [
        'query' => [
            'match_phrase' => [
                'title' => [
                    'query' => '开源技术小栈20245直播间',
                    'slop' => 2
                ]
            ]
        ]
    ]
];
$indexResponse = $client->search($indexParams);

Multi‑field Search (multi_match)

The multi_match query lets a single query string be searched across multiple fields such as title and content. The simplest form specifies the query and the target fields.

$client = ElasticSearchCLient::getInstance();
$query = '开源技术小栈';
$indexParams = [
    'index' => 'resty_product_test_index',
    'body' => [
        'query' => [
            'multi_match' => [
                'query' => $query,
                'fields' => ['title', 'content']
            ]
        ]
    ]
];
$indexResponse = $client->search($indexParams);

Results contain hits where either title or content includes the phrase "开源技术小栈".

The type parameter changes the matching algorithm. For example, using "best_fields" selects the single field with the highest score.

$client = ElasticSearchCLient::getInstance();
$query = '开发者社区';
$indexParams = [
    'index' => 'resty_product_test_index',
    'body' => [
        'query' => [
            'multi_match' => [
                'query' => $query,
                'fields' => ['title', 'content'],
                "type" => "best_fields"
            ]
        ]
    ]
];
$indexResponse = $client->search($indexParams);

This query searches the same two fields but applies the "best_fields" strategy, which prefers the field that yields the highest relevance score for each document.

Advanced Query String

The query_string query supports Lucene syntax, enabling complex logical operators and field‑specific searches. A basic example searches the title field for the keyword "开发者社区".

$client = ElasticSearchCLient::getInstance();
$query = '开发者社区';
$indexParams = [
    'index' => 'resty_product_test_index',
    'body' => [
        'query' => [
            'query_string' => [
                'default_field' => 'title',
                'query' => $query
            ]
        ]
    ]
];
$indexResponse = $client->search($indexParams);

This query returns documents whose title contains the term "开发者社区" while allowing the full expressive power of Lucene query syntax for more sophisticated searches.