Start your content search
13 min read

Search Type features

Search type features modify how the search engine match works. Empathy Platform includes the following search type features:

  • Match All Search: Performs a search to match all documents in the index.
  • Match Search: Performs a search on a single field in the index.
  • Multi Match Search: Performs a search on a collection of fields in the index.
  • Term Search: Performs a search using an exact match query on a single field in the index.
  • Browsing Search: Performs navigation using a specific value for the field, filtering results, and applying faceting and boosting configuration for the field value.
  • Top Clicked Search: Performs a search for the most clicked products in the product catalogue.
  • Empathize Match Search: Performs a search for query suggestions and popular searches.
  • Field Values: Lists all possible values for a field.

warning

You can only use one query type feature at the same time.

interact

For an explanation of how to set up the features in the Search microservice, see Feature configuration.

The Match All Search (matchAllSearch) feature lets you add a query to match all fields from all documents.

Search configuration

There are no additional parameters to configure this feature.

Configuration example

{
  "name": "matchAllSearch"
}

The Match Search (matchSearch) feature lets perform a search engine match by using the configured field parameter. If the query parameter is not present in the request, it applies matchAllSearch behavior, if configured. Otherwise, it gives an exception and requests the missing query parameter.

Field Description Type Default value
query Query terms to search (optional) String -

Query can be sent as a parameter with the format query=sample query. The feature uses the field name provided within the value specified through the query parameter query if present. If the query parameter is not present within the request and emptyQueryParamAllowed is false or undefined, the request gives an 400 error.

Search configuration

You configure the fields to search on in the JSON configuration file of the feature, not in the request.

Config parameters

Field Description Type Default value
field Field to be used for applying the match (required) String ---
queryOperator Query operator to be used. Available values: AND, OR (optional) String OR
emptyQueryParamAllowed Determines whether a query parameter is required or not.
If true, it allows requests without the query parameter (optional)		 Boolean False

Configuration example

{
  "name": "matchSearch",
  "config": {
    "field": "keywords",
    "queryOperator": "AND",
    "emptyQueryParamAllowed": true
  }
}

The Multi Match Search (multiMatchSearch) feature adds the ability to search on a collection of fields and returns the associated results.

Multi Match Search performs two functions:

  • Query: It performs a search for the query term on a collection of fields.
  • Equalize: It retrieves the Equalize configuration, which contains the relationship between the fields and their weighting, and adds the weights to each field so that they can be more or less significant for the query.

Configuration for query usage

Fields to search on are not configured through the request but in the JSON configuration file of the feature.

Config parameters

Field Description Type Default value
searchableFields List of allowed fields to be used to search over them. It cannot be empty or null (required) Array ---
tieBreaker Capability to resolve conflicts resulting from identical product overall scores (optional) Integer 0.0
autoGenerateSynonymsPhraseQuery Specify if match phrase queries must be created for multi-term synonyms (optional) Boolean false

Configuration for Equalize

The Equalize configuration can be retrieved via JSON configuration (only for EQ Default config), via the Playboard, or by request.

EQ Default JSON configuration

The EQ Default configuration settings are established for each field via a JSON configuration in the Instance Management console. The fields that can be weighted are also handled via JSON.

The weights for the fields are calculated based on the type and the Equalize boost using the formula: field_type_multiplier * equalize_field_boost.

The field subtypes are:

Type Description Multiplier
Keyword Field with exact match 2.0
Text Field using standard tokenizer (opens new window), lowercase filter, and ASCII folding 1.0
NGram Field with text abilities and ngram tokenizer (opens new window) 0.05
Derived Field with text abilities and stemming (opens new window) 0.2

Config parameters

Field Description Type Default value
searchableFields List of allowed Equalized fields to be used to search over them. It cannot be empty or null (required) Array ---
tieBreaker Behavior to calculate the score (optional). Available values are detailed in a note below. Integer 0.0
autoGenerateSynonymsPhraseQuery Specify if match phrase queries must be created for multi-term synonyms (optional) Boolean false
multipliers Overrides the search multipliers. It can only be defined for the default configuration. See the table below for more information about the available values --- false

Tiebreaker values

  • 0.0: Take the single best score out of all the fields (for example, first_name:will and last_name:will).
  • 1.0: Add together the scores for all the fields (for example, first_name:will and last_name:will).
  • 0.0 < n < 1.0: Take the single best score plus tie-breaker, multiplied by each of the scores from other matching fields/groups.

Multipliers config

Field Description Mandatory
keyword Multiplier for exact match yes
text Multiplier for text match yes
derived Multiplier for synonym matches yes
ngram Multiplier for partial word match yes
useSimplifiedFieldMultiplier In case text and derived multipliers are the same, the feature will only include derived value, as both will contain the same transformations no
useCompleteWords Only takes into account complete words no

Configuration example

      {
      "name": "multiMatchSearch",
      "config": {
        "searchableFields": [
          {
            "name": "description",
            "boost": 3
          },
          {
            "name": "brand"
          },
          {
            "name": "category",
            "boost": 0
          }
        ],
        "tieBreaker": 0.1,
        "autoGenerateSynonymsPhraseQuery": false,
        "multipliers": {
          "keyword": 2.0,
          "text": 1.0,
          "derived": 0.2,
          "ngram": 0.05,
          "useSimplifiedFieldMultiplier": true
        }
      }
    }

Equalize configurations via the Playboard

Tailored Equalize configurations are established using the Equalize Management tool in the Empathy Platform Playboard. Weights applied are retrieved from the tool and each Equalize configuration, the EQ Default included as well, is assigned with an equalizeId that is displayed in the configuration list of the Equalize management tool.

Equalize configurations retrieved via request

The Equalize configurations can be retrieved via request. An Equalize configuration is sent as a query parameter with the format equalize={"field1":value, "field2":value}. The value of the parameter must be a JSON object that contains a collection of fields, and the related weight.

  • Request by ID: You can additionally send a request with the parameter equalizeId indicating the id (UUID) of a specific configuration to be used, having priority and being used over any other equalizes in the string. When the indicated id has an invalid format or a configuration with that id doesn't exist, a 400 Bad Request error will be returned.

    Equalize configuration ID

    Since all the tooling features in Empathy are A/B testing-ready. You to create different relevancy models for different segments in your commerce store and easily test their performance with your favorite A/B testing tool.

The Term Search (termSearch) feature adds the ability to search using an exact match query on a single field and returns the associated results.

Field Description Type Default value
query Query term to search (required) String ---

The query must be sent as a parameter with the format query=term.

Search configuration

The field to search is configured in the JSON configuration file of the feature.

Config parameters

Field Description Type Default value
field Name of the field to be used to search. It cannot be empty or null (required) String ---

Configuration example

{
  "name": "termSearch",
  "config": {
    "field": "color"
  }
}

The Browsing Search (browsingSearch) feature adds the ability to navigate based on a specific value for a field. Navigating means not only filtering the results and all facets, but also applying specific faceting and boosting configuration for the selected field value.

You must send the browseField and browseValue parameters with the field and the value associated to filter the results.

Parameters must be sent with the format browseField=field&browseValue=value.

Field Description Type Default value
browseField Field to filter the results (required) String ---
browseValue Value associated to the field to filter the results (required) String ---

Search configuration

There are no additional parameters to configure this feature.

Configuration example

{
  "name": "browsingSearch"
}

The Top Clicked Search (topClickedSearch) feature lets you query the most clicked products in the catalogue. Once configured, the filters listed in the configuration can be used as query parameters.

For example, if the feature is configured with the filters lang and scope, the query parameters are: lang=en&scope=mobile.

When a filter is configured, only the products that strictly match the corresponding query parameter are returned. This means that:

  • If there is a query parameter, the products matching the query parameter are returned.
  • If there is NO query parameter, only the products without a value for that filter are returned.

You can enable a Redis cache to cache the results of the Top Clicked query.

Search configuration

Config parameters

Field Description Type Default value
idFieldName Name of the field used as productId in the product catalogue (required) String ---
filters Filters to be applied when querying the top-clicked index (optional) Array ---
redisCached Enable Redis cache (optional) Boolean ---
redisTTL Time-to-live for the data in Redis cache, in seconds (optional) Integer ---

Filters parameters

Field Description Type Default value
queryParam Name of the query parameter (required) String ---
field Name of the field in the top-clicked index (optional) Array Value set for queryParam

Configuration example

{
  "name": "topClickedSearch",
  "config": {
    "idFieldName": "groupId",
    "filters": [
      {
        "queryParam": "scope",
        "field": "filters.scope"
      }
    ],
    "redisCached": true,
    "redisTTL": 250
  }
}

The Empathize Match Search (empathizeMatchSearch) feature lets you run an Elastic match by using the configured field and applying a matchAll if the query parameter is not present within the request and the configuration allows that behavior. Otherwise, it returns an exception asking for a missing query parameter.

Field Description Type Default value
query Query terms to search (optional) String ---

The query can be sent as a parameter with the format query=sample query. The feature uses the field name provided within the value specified through the query parameter query if present. If the query parameter is not present within the request and emptyQueryParamAllowed is false or undefined, the request gives a 400 error.

Search configuration

The field to search is configured in the JSON configuration file of the feature.

Config parameters

Field Description Type Default value
field Field to be used for applying the match (required) String ---
queryOperator Query operator to be used. Allowed operators: AND, OR (optional) String OR
emptyQueryParamAllowed Determines whether empty query parameters are allowed. If true, it allows requests without query parameter (optional) Boolean false
excludeExactMatch Determines whether exact matches are allowed. If true, it removes the results that exactly match the query (optional) Boolean false

Configuration example

{
  "name": "empathizeMatchSearch",
  "config": {
    "field": "keywords",
    "queryOperator": "AND",
    "emptyQueryParamAllowed": true,
    "excludeExactMatch": false
  }
}

Field Values

The Field Values (fieldvalues) feature lets you list all the possible values for a field. The accepted fields can be configured in two ways:

  • Fields configured as allowedFields in the client.json file.
  • Attribute fields can be informed as query parameters, but they must be a subset of the fields included in the configuration.

The endpoint accepts a query parameter attribute in the request. Where the query parameter is informed, an aggregation for that field is included in the response. Only one field is accepted.

Example:

attribute=fieldName

Search configuration

You configure the fields to be used in the JSON configuration.

Config parameters

Field Description Type Default value
allowedFields Fields that can be sent as parameters for this feature (required) String ---
field Field name (required) String ---
aggregationSize Maximum size the aggregation should have (required) Integer ---
aggregationField Elasticsearch property by which the aggregations should be made (required) String ---

Configuration example

{
  "name": "fieldValues",
  "config": {
    "allowedFields": [
      {
        "field": "brand",
        "aggregationSize": 21000,
        "aggregationField": "facet_brandName"
      }
    ]
  }
}
Cookie-Less