Boost

Boosting increases the importance of selected matches so they receive higher relevance scores and tend to appear earlier in Elasticsearch results. It is useful when a search spans several fields but a match in one field matters more to users—for example, a book title should usually outweigh the same words appearing only in a synopsis or tags. Full-text hits carry a positive floating-point `_score`; higher scores rank first. Elasticsearch calculates these relevance scores with BM25, then a boost adjusts the weight given to chosen criteria. section 2.3.4

For a `multi_match` query, append `^` and a numeric factor to a field name. `title^3` means that a title match is weighted three times as strongly as a synopsis match. This does not restrict the search to titles: both fields remain searchable. It changes the ordering when otherwise relevant documents compete. section 2.3.5 section 10.8.4

GET books/_search
{
  "query": {
    "multi_match": {                      #1
      "query": "Java",
      "fields": ["title^3","synopsis"]    #2
    }
  }
}
#1 Searches through multiple fields
#2 A caret followed by the boost number indicates the field being boosted.

Use this request as the basic pattern: search `title` and `synopsis`, but favor documents whose title matches `Java`. The caret and number belong in the field entry, not beside the query text. section 2.3.5

Boosting can also assign a score at the query level. A plain `match_all` query gives every returned document a default score of `1.0`; adding `"boost": 2` raises that score. This is appropriate when all documents matched by a query need the same predefined score, rather than when one text field should outweigh another. section 10.3.1

Filters normally run without a relevance score. Wrap a filter in `constant_score` and supply `boost` to give every filtered result the same nonzero score. For example, a rating-range filter can receive score `5.0`. Combined with a `match` query in a `bool` query, a `constant_score` clause can promote all black TVs by `3.5` while still requiring a product match. section 11.4

A `bool` query can boost results through its `should` clauses as well. Documents must satisfy the `must` criteria, while matching optional `should` criteria adds weight to their relevance score. In the TV example, an LG TV matching price or “4K Ultra HD” preferences rises from a score of `4.4325914` to `14.9038105`. Use `minimum_should_match` when optional criteria must meet a minimum count. section 11.3.6

Choose the boosting form that matches the ranking decision:

• Use `field^factor` in `multi_match` when one searched field, such as `title`, is more meaningful than another. section 2.3.5
• Use a query-level `boost` when the entire query needs a predefined score, including `match_all`. section 10.3.1
• Use `constant_score` when a filter needs a fixed score, or `should` clauses when optional matches should add relevance. section 11.4 section 11.3.6