Building Search API with ElasticSearch — Part 06: Filters

3 min readJul 11, 2022

In the previous post we created facets to refine the search results and to finish this work we will use the filters.

The filter is a clause that will say whether or not the document matches the criterion and it does not affect the document’s score.
Another important point is that the filters are cached and this optimizes the query performance.

Java Code

To create the filters, we will define in the SearchFilter the fields that will receive the parameters. The fields selected for filtering will be “genre” and “certificate”.

These filters became a Map of filters because we will pass this Map to the SearchRequest.
To add the filter it is necessary to use the Bool Query and it will be in this clause that we will use the Terms Query because we will use a list of terms for filtering.

Running test

The test will be with the term “batman”, the result will be a list with 4 movies.

curl --location --request GET 'http://localhost:8080/api/movies/search?size=10&text=batman'

Results:

{
    "size": 10,
    "total": 4,
    "suggestion": "",
    "movies": [
        {
            "title": "Batman Begins",
            "certificate": "UA",
            "genre": [
                "Action",
                "Adventure"
            ],
            "search_after": [
                94.20764,
                156
            ]
        },
        {
            "title": "Batman: Mask of the Phantasm",
            "certificate": "PG",
            "genre": [
                "Animation",
                "Action",
                "Crime"
            ],
            "search_after": [
                79.15011,
                666
            ]
        },
        {
            "title": "The Dark Knight",
            "certificate": "UA",
            "genre": [
                "Action",
                "Crime",
                "Drama"
            ],
            "search_after": [
                9.961502,
                3
            ]
        },
        {
            "title": "The Dark Knight Rises",
            "certificate": "UA",
            "genre": [
                "Action",
                "Adventure"
            ],
            "search_after": [
                9.7500925,
                64
            ]
        }
    ]
}

Executing the query that returns the facets we have the result below:

curl --location --request GET 'http://localhost:8080/api/movies/facets?text=batman'

We want to refine the results, we select the genre “Crime”, we hope to have only two documents as results.

curl --location --request GET 'http://localhost:8080/api/movies/search?size=10&text=batman&genres=Crime'

Results:

Now let’s filter again by the “PAG” certificate.

curl --location --request GET 'http://localhost:8080/api/movies/search?size=10&text=batman&genres=Crime&certificates=PG'

And the result will be just 1 document.

We finish here, in the next article we will implement a term correction feature known as “did you mean”.

References

Query and filter context | Elasticsearch Guide [8.3] | Elastic

The relevance score is a positive floating point number, returned in the _score metadata field of the search API. The…

www.elastic.co

Terms query | Elasticsearch Guide [8.3] | Elastic

Returns documents that contain one or more exact terms in a provided field. The terms query is the same as the , except…