Home / Documentation / REST APIs / Resolution API

Resolution API

Runs an entity resolution job and returns the results.

The request accepts two endpoints:

POST _zentity/resolution
POST _zentity/resolution/{entity_type}

Example Request

This example request resolves a person identified by a name, a dob, and two phone values, while limiting the search to one index called users_index and two resolvers called name_dob and name_phone. The request passes a param called fuzziness to the phone attribute, which can be referenced in any matcher clause that uses the fuzziness param. Note that an attribute can accept either an array of values or an object with the values specified in a field called "values". It's also valid to specify an attribute with no values but to override the default params, such as to format the results of any date attributes in the response.

Read the input specification for complete details about the structure of a request.

POST _zentity/resolution/person?pretty
{
  "attributes": {
    "name": [ "Alice Jones" ],
    "dob": {
      "values": [ "1984-01-01" ]
    },
    "phone": {
      "values": [
        "555-123-4567",
        "555-987-6543"
      ],
      "params": {
        "fuzziness": 2
      }
    }
  },
  "scope": {
    "exclude": {
      "attributes": {
        "name": [
          "unknown",
          "n/a"
        ],
        "phone": "555-555-5555"
      }
    },
    "include": {
      "indices": [
        "users_index"
      ],
      "resolvers": [
        "name_dob",
        "name_phone"
      ]
    }
  }
}

Example Response

This example response took 64 milliseconds and returned 2 hits. The _source field contains the fields and values as they exist in the document indexed in Elasticsearch. The _attributes field contains any values from the _source field that can be mapped to the "attributes" field of the entity model. The _hop field shows the level of recursion at which the document was fetched. Entities with many documents can span many hops if they have highly varied attribute values.

Read the output specification for complete details about the structure of a response.

{
  "took": 64,
  "hits": {
    "total": 2,
    "hits": [
      {
        "_index": "users_index",
        "_id": "iaCn-mABDJZDR09hUNon",
        "_hop": 0,
        "_attributes": {
          "city": "Beverly Halls",
          "first_name": "Alice",
          "last_name": "Jones",
          "phone": "555 123 4567",
          "state": "CA",
          "street": "123 Main St",
          "zip": "90210-0000"
        },
        "_source": {
          "@version": "1",
          "city": "Beverly Halls",
          "fname": "Alice",
          "lname": "Jones",
          "phone": "555 987 6543",
          "state": "CA",
          "street": "123 Main St",
          "zip": "90210-0000"
        }
      },
      {
        "_index": "users_index",
        "_id": "iqCn-mABDJZDR09hUNoo",
        "_hop": 0,
        "_attributes": {
          "city": "Beverly Hills",
          "first_name": "Alice",
          "last_name": "Jones",
          "phone": "(555)-987-6543",
          "state": "CA",
          "street": "123 W Main Street",
          "zip": "90210"
        }
        "_source": {
          "@version": "1",
          "city": "Beverly Hills",
          "fname": "Alice",
          "lname": "Jones",
          "phone": "(555)-987-6543",
          "state": "CA",
          "street": "123 W Main Street",
          "zip": "90210"
        }
      }
    ]
  }
}

HTTP Headers

URL Parameters

URL Parameters (advanced)

These are advanced search optimizations. Most users will not require them. It's recommended to use the default settings of the cluster unless you know what you're doing.

Request Body Parameters

Notes

• If you define an entity_type, zentity will use its model from the .zentity-models index.
• If you don't define an entity_type, then you must include a model object in the request body.
• You can define an entity_type in the request body or the URL, but not both.

Tips

• If you only need to search a few indices, use scope.exclude.indices and scope.include.indices parameter to prevent the job from searching unnecessary indices in the entity model at each hop.
• Beware if your data is transactional or has many duplicates. You might need to lower the values of max_hops and max_docs_per_query if your jobs are timing out.
• Use scope.exclude.attributes to prevent entities from being over-resolved (a.k.a. "snowballed") due to common meaningless values such as "unknown" or "n/a".
• Use scope.include.attributes to limit the job within a particular context, such as by matching documents only within a given state or country.

Continue Reading