如何修改Elasticsearch查询以实现entity_type字段精确匹配、predicates.name.value字段部分匹配并优先排序精确匹配结果

Got it, let's tweak your query to hit all your requirements: exact match on entity_type, partial match on predicates.name.value, and ensuring exact matches on the latter field show up first. Here's how to do it:

Key Changes Needed: #

• Exact match for entity_type: Use a term query instead of match—match is built for full-text search (it analyzes input), while term does strict exact matching (ideal for fields like entity types that are typically keyword-typed).
• Partial + prioritized exact match for predicates.name.value: Add a should clause inside your bool query to include both an exact match (with a score boost) and a partial match. This way, results with exact matches get a higher score and appear at the top, while partial matches still get included in the results.
• Enforce minimum match requirement: Set minimum_should_match: 1 to ensure at least one of the should clauses (exact or partial) is satisfied alongside the mandatory entity_type match.

Modified Query Code: #

self.search = {
  "query": {
    "bool": {
      "must": [
        { "term": { "entity_type": entity_type } } // Strict exact match on entity_type
      ],
      "should": [
        { 
          "term": { 
            "predicates.name.value": { 
              "value": criteria, 
              "boost": 10 // Boost exact matches to rank them higher
            } 
          }
        },
        { "match": { "predicates.name.value": criteria } } // Partial full-text match
      ],
      "minimum_should_match": 1 // Require at least one should clause to be met
    }
  }
}

Quick Notes: #

• If predicates.name.value is a text field (not a keyword field), you’ll need to use predicates.name.value.keyword in the term query for exact matching—text fields are analyzed, so their raw unanalyzed values are stored in the keyword sub-field by default.
• If you need wildcard-style partial matching (like *criteria* or criteria*), replace the match clause with a wildcard query instead:

  { "wildcard": { "predicates.name.value": "*" + criteria + "*" } }
  

  Just keep in mind wildcard queries can be slower than match queries, so use them only if full-text partial matching doesn’t fit your needs.

内容的提问来源于stack exchange，提问作者Greg Finzer