Query Vs Filter in Elastic Search

Question

I am trying to index a document which has three fields first_name, last_name, occupation of type "keyword" and has values XYZ, ABC, DEF respectively.

I have written query using filter for an exact match with AND condition as follows,

"query": {
  "bool": {
    "filter": [
      {"term": {"first_name": "XYZ"}},
      {"term": {"last_name": "ABC"}}
    ]
  }
}

This has to return one document, but returns nothing.

I have another query for the same operation,

"query": {
  "bool": {
    "must": [
      {"match": {"first_name": "XYZ"}},
      {"match": {"last_name": "ABC"}}
    ]
  }
}

This returns one document.

According to Elasticsearch documentation, I understand that the difference between query and filter is that filter does not score the result. I am not sure why the first query does not return any result. Is my understanding correct?

Answer 1

As documentation states there is no difference between query and filter except scoring. Of course this applies to the situation when both query and filters uses the same query type. Here you are using two different types - term and match. term is designed for exact comparison while match is analyzed and used as full-text search.

Take a look at the example below.

Your mapping:

PUT /index_53053054
{
  "mappings": {
    "_doc": {
      "properties": {
        "first_name": {
          "type": "text"
        },
        "last_name": {
          "type": "text"
        },
        "occupation": {
          "type": "keyword"
        }
      }
    }
  }
}

Your document:

PUT index_53053054/_doc/1
{
  "first_name": "XYZ",
  "last_name": "ABC",
  "occupation": "DEF"
}

filter query:

GET index_53053054/_search
{
  "query": {
    "bool": {
      "filter": [
        {
          "match": {
            "first_name": "XYZ"
          }
        },
        {
          "match": {
            "last_name": "ABC"
          }
        },
        {
          "term": {
            "occupation": "DEF"
          }
        }
      ]
    }
  }
}

and result:

{
  "took": 7,
  "timed_out": false,
  "_shards": {
    "total": 5,
    "successful": 5,
    "skipped": 0,
    "failed": 0
  },
  "hits": {
    "total": 1,
    "max_score": 0,
    "hits": [
      {
        "_index": "index_53053054",
        "_type": "_doc",
        "_id": "1",
        "_score": 0,
        "_source": {
          "first_name": "XYZ",
          "last_name": "ABC",
          "occupation": "DEF"
        }
      }
    ]
  }
}

Similar must query:

GET index_53053054/_search
{
  "query": {
    "bool": {
      "must": [
        {
          "match": {
            "first_name": "XYZ"
          }
        },
        {
          "match": {
            "last_name": "ABC"
          }
        },
        {
          "term": {
            "occupation": "DEF"
          }
        }
      ]
    }
  }
}

and response:

{
  "took": 5,
  "timed_out": false,
  "_shards": {
    "total": 5,
    "successful": 5,
    "skipped": 0,
    "failed": 0
  },
  "hits": {
    "total": 1,
    "max_score": 0.8630463,
    "hits": [
      {
        "_index": "index_53053054",
        "_type": "_doc",
        "_id": "1",
        "_score": 0.8630463,
        "_source": {
          "first_name": "XYZ",
          "last_name": "ABC",
          "occupation": "DEF"
        }
      }
    ]
  }
}

As you can see hits are almost the same. The only difference is that in filter score is not calculated while in must query is.

Read more: https://www.elastic.co/guide/en/elasticsearch/reference/6.4/query-filter-context.html