Sometimes it is useful to obtain only records that have non-null and non-empty values in a certain field. For example, getting all products that have description fields defined:
GET /amazon_products/_search
{
"query": {
"exists": {
"field": "description"
}
}
}
The exists query turns the query into a filter; in other words, it runs in a filter context. This is similar to the range query where the scores don't matter.
What is a Filter Context? When the query is just about filtering our documents, that is, deciding whether to include the document in the result or not, it is sufficient to skip the scoring process. Elasticsearch can skip the scoring process for certain types of queries and assign a uniform score of 1 to each document, which passes the filter criteria. This not only speeds up the query (as the scoring process is skipped) but also allows Elasticsearch to cache the results of filters. Elasticsearch caches the results of filters by maintaining arrays of zeros and ones.