패싯 검색을 위한 집계

Search finds documents; aggregations summarize them. They power faceted navigation, dashboards, and analytics, all in the same request as the search.

Aggregations come in two main families:

• Bucket aggregations group documents (e.g. by category)
• Metric aggregations compute numbers (e.g. average price)

The terms aggregation produces a bucket per distinct value with counts, the classic facet.

Setting size: 0 returns no document hits, only the aggregation results, which is efficient when you only want the facets.

Aggregations run on non-analyzed fields. Use the keyword field, not the analyzed text field, or you will bucket on individual tokens.

Metrics like avg, min, max, and sum compute a single value over matching documents.

A histogram buckets numeric values into fixed intervals, great for price ranges.

A date_histogram groups by calendar intervals like day or month, ideal for time-series charts.

You can nest a metric inside a bucket: average price within each category. Nesting is how rich facets are built.

Aggregations are computed over the documents that match the query and filters, so facet counts naturally reflect the current search context.

The cardinality metric gives an approximate distinct-value count, far cheaper than exact counting on large data.

You learned aggregations for faceted search: bucket aggregations like terms and histograms, metric aggregations like avg and cardinality, nesting metrics inside buckets, the size: 0 trick, and why facets use keyword fields.